NEWS.md 99.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
Announcement for Dynare 4.6.1 (on 2020-03-13)
=============================================

We are pleased to announce the release of Dynare 4.6.1.

This maintenance release fixes various bugs.

The Windows, macOS and source packages are already available for download at
[the Dynare website](https://www.dynare.org/download/).

All users are strongly encouraged to upgrade.

This release is compatible with MATLAB versions ranging from 7.9 (R2009b) to
9.7 (R2019b), and with GNU Octave versions 5.2.0 (under Windows) and 4.4.1
(under macOS).

Here is a list of the problems identified in version 4.6.0 and that have been
fixed in version 4.6.1:

* Installation on macOS would fail if the GCC compiler was supposed to be
  installed and `www.google.com` was not reachable or blocked
* Dynare++ was missing the `dynare_simul.m` file
* The parameter vector `M_.params` would not be correctly updated after calls
  to `stoch_simul` and `discretionary_policy` if parameters had been modified
  in a steady state file
* The `stoch_simul` command with both the `nograph` and `TeX` options would
  crash
* The `stoch_simul` command with the `noprint` option would crash
* The `prior moments` command would crash if the used parameter vector
  triggered an error code
* In case of problem, the `discretionary_policy` command would crash instead of
  aborting with a proper error code
* Computing of prior/posterior statistics would not work in parallel
* Closing of parallel estimation on GNU/Linux could crash
* The `histval` command would not work in combination with the
  `predetermined_variables` command
* Ramsey optimal policy with multiple instruments would crash if a steady state
  file returned complex values, instead of providing an error message
* The `model_diagnostics` command would not correctly update the parameter
  vector if the latter was set in a steady state file
* The `model_diagnostics` command would ignore the `nocheck` steady state flag


44 45
Announcement for Dynare 4.6.0 (on 2020-02-20)
=============================================
46 47 48 49 50 51 52 53 54 55 56 57 58 59

We are pleased to announce the release of Dynare 4.6.0.

This major release adds new features and fixes various bugs.

The Windows, macOS and source packages are already available for download at
[the Dynare website](https://www.dynare.org/download/).

All users are strongly encouraged to upgrade.

This release is compatible with MATLAB versions ranging from 7.9 (R2009b) to
9.7 (R2019b), and with GNU Octave versions 5.2.0 (under Windows) and 4.4.1
(under macOS).

60 61
Major user-visible changes
--------------------------
62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208

 - Stochastic simulations

    - The perturbation method is now available at an arbitrary approximation
      order. In other words, the `order` option of `stoch_simul` accepts an
      arbitrary positive integer (of course, up to some model-specific
      computational limit).

    - New option `filtered_theoretical_moments_grid` of `stoch_simul`, that
      supersedes `hp_ngrid`.

 - Estimation

    - Nonlinear estimation is now also available at an arbitrary approximation
      order. In other words, the `order` option of `estimation` accepts an
      arbitrary positive integer (of course, up to some model-specific
      computational limit).

    - Various improvements to particle filters.

    - It is now possible to estimate models under optimal policy (see below).

    - Variance decomposition of observables now accounts for measurement error.

    - New option `mh_tune_jscale` of `estimation` command for tuning the scale
      parameter of the proposal distribution of the Random Walk Metropolis
      Hastings.

    - Added debugging info when parameters take a `NaN` or `Inf` value.

    - Option `mode_compute=1` is now available under Octave.

 - Perfect foresight and extended path

    - A significant speed improvement should be noted on large models (when
      neither `bytecode` nor `block` option is used). The stacked problem is
      now constructed using a dedicated machine-compiled library that greatly
      speeds up the process (in particular, the time spent in that step can
      become negligible when the `use_dll` option is used).

    - New options `print` and `noprint` of `perfect_foresight_solver` command.

    - Option `stack_solve_algo=2` is now available under Octave.

 - Steady state

    - Option `solve_algo=7` is now available under Octave.

 - Optimal policy

    - The `ramsey_policy` command is now deprecated. It is superseded by
      successive calls to `ramsey_model`, `stoch_simul`, and
      `evaluate_planner_objective` (in this order).

    - It is now possible to estimate a model under optimal policy (either
      Ramsey or discretionary) by running the `estimation` command after either
      `ramsey_model` or `discretionary_policy`. It is however not yet possible
      to estimate parameters that appear in the discount factor of the social
      planner.

    - Discretionary policy returns a more informative error message when the
      objective has nonzero derivatives with respect to some variables.

 - Identification

    - Added minimal system identification check of *Komunjer and Ng (2011)*.

    - Added spectrum identification check of *Qu and Tkachenko (2012)*.

    - Identification is now also available for approximation orders 2 and 3
      with either analytical or numerical parameter derivatives. The relevant
      moments and spectrum are computed from the pruned state space system
      as in *Mutschler (2015)*.

    - All tests (moments, spectrum, minimal system, strength) can be turned
      off.

    - More numerical options can be changed by the user.

    - Improved printing and storage (same folder) of results.

 - Sensitivity analysis

    - New `diffuse_filter` option to the `dynare_sensitivity` command.

    - Arbitrary expressions can now be passed for the interval boundaries in
      `irf_calibration` and `moment_calibration`. ⚠ This breaks the
      previous syntax, requiring that the lower/upper bounds be separated by
      commas.

 - Forecasting and smoothing

    - In `conditional_forecast_paths`, it is no longer required that all
      constrained paths be of the same length. There may now be a different
      number of controlled variables at each period. In that case, the order of
      declaration of endogenous controlled variables and of `controlled_varexo`
      matters: if the second endogenous variable is controlled for less periods
      than the first one, the second `controlled_varexo` isn't set for the last
      periods.

    - New option `parameter_set` to the `calib_smoother` command.

    - ⚠ The results of `conditional_forecast` command is now saved in
      `oo_` (used to be in a file)

 - Shock decomposition

   - Added `fast_realtime` option to real time shock decomposition (deactivated
     by default, runs the smoother only twice: once for the last in-sample and
     once for the last out-of-sample data point).

   - New `diff`, `flip`, `max_nrows`, `plot_init_date` and `plot_end_date`
     options to `plot_shock_decomposition`.

   - New `initial_decomposition_decomposition` command, for computing and
     plotting the decomposition of the effect of smoothed initial conditions of
     state variables.

   - New `squeeze_shock_decomposition` command, for removing decompositions of
     variables that are not of interest.

   - New `with_epilogue` option (common to `shock_decomposition`,
     `realtime_shock_decomposition` and `initial_condition_decomposition`).

   - New `init2shocks` block to attribute initial conditions to shocks.

 - Macro processor

   - New object types: real (supersedes integers), boolean (distinct from
     integers), tuple, user-defined function.

   - New operators: various mathematical functions, set operations on arrays
     (union, intersection, difference, cartesian power and product), type
     checking and conversion.

   - Added support for comprehensions (*e.g.* the set containing the squares of
     all even numbers between 1 and 5 can be constructed with `[ i^2 for i in
     1:5 when mod(i,2) == 0]`).

   - User-defined functions can be declared using the `@#define` operator (*e.g.*
     `@#define f(x) = 2*x^2+3*x+5`).

   - `@#elseif`-clauses are now supported in conditional statements.

   - `@#for` loops can iterate over several variables at the same time (*e.g.*
     `@#for (i,j) in X`, where `X` is an array containing tuples of size 2).

209 210 211 212
   - Added the possibility to exclude some elements when iterating over `@#for`
     loops (*e.g.* `@#for i in 1:5 when mod(i,2) == 0` iterates over all even
     numbers between 1 and 5).

213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
   - A `defined()` function allows testing whether macro variables have been
     defined.

   - Empty arrays (with the `[]` syntax) are now possible.

   - Arrays of arrays are now supported.

   - New macro directives `@#echomacrovars` and `@#echomacrovars(save)` for
     displaying or saving the values of all macro-variables.

   - Inline comments are now supported.

   - ⚠ All division operations are now done with doubles (as opposed to
     integers). To achieve the old functionality, use the new `floor` operator.

   - ⚠ Colon syntax used to require braces around it to create an array
     (*e.g.* `[1:3]` would create `[1,2,3]`). Now this is not necessary (`1:3`
     creates `[1,2,3]` while `[1:3]` would create `[[1,2,3]]`).

   - ⚠ Previously, printing a boolean would print `1` or `0`. Now, it
     prints `true` or `false`. To achieve the old functionality, you must cast
     it to a real, *e.g.* `@{(real)(1!=0)}`.

 - LaTeX output

    - New command `write_latex_steady_state_model`.

    - New option `planner_discount_latex_name` of `ramsey_model` and
      `discretionary_policy`.

    - New command `model_local_variable` command for assigning a LaTeX name to
      model-local variables.

    - The `write_latex_static_model` and `write_latex_original_model` commands
      now support the `write_equation_tags` option.

 - Compilation of the model (`use_dll` option) made easier and faster

    - Under Windows, it is no longer necessary to manually install the
      compiler, since the latter is now shipped by the Dynare installer.

    - Under macOS, the Dynare installer now automatically downloads and
      installs the compiler.

    - It is no longer necessary to configure MATLAB to let it know where the
      compiler is, since the compilation is now done by the preprocessor.

    - The compilation phase is now faster on large models (this has been
      achieved by disabling a few time-consuming and not-so-useful optimization
      passes otherwise done by the compiler).

    - New `compilation_setup` block for specifying a custom compiler or custom
      compilation flags.

 - Model, variables and parameters declaration

    - New syntax to declare model variables and parameters on-the-fly in the
      `model` block. To do this, simply follow the symbol name with a vertical
      line (`|`, pipe character) and either an `e`, an `x`, or a `p`. For
      example, to declare a parameter named `alpha` in the model block, you
      could write `alpha|p` directly in an equation where it appears.
      Similarly, to declare an endogenous variable `c` in the model block you
      could write `c|e`.

    - New syntax to declare model variable and parameters on-the-fly in
      equation tags. In the tag, simply state the type of variable to be
      declared (`endogenous`, `exogenous`, or `parameter` followed by an equal
      sign and the variable name in single quotes. Hence, to declare a variable
      `c` as endogenous in an equation tag, you can type `[endogenous='c']`.

    - New `epilogue` block for computing output variables of interest that may
      not be necessarily defined in the model (*e.g.* various kinds of
      real/nominal shares or relative prices, or annualized variables out of a
      quarterly model).

 - Command-line options

    - Added the possibility to declare Dynare command-line options in the `.mod`
      file.

    - New option `nopreprocessoroutput` to disable printing of messages from
      the preprocessor.

    - It is now possible to assign an arbitrary macro-expression to a
      macro-variable defined on the command-line, using the `-D` syntax.

    - New  option `linemacro` to revert to the old format of the
      macro-processed file (see below).

 - Preprocessor outputs and inputs

    - Added JSON output to the preprocessor. A representation of the model file
      and the whole content of the `.mod` file is saved in `.json` files.
      These JSON files can be easily parsed from any language (C++, Fortran,
      Python, Julia, MATLAB, Octave…). This new feature opens the possibility to
      develop alternative back-ends for the Dynare language.

    - ⚠ Most files generated by the preprocessor are now grouped under
      two subdirectories. Assuming your file is `FILENAME.mod`, then M-files
      and MEX-files will be under `+FILENAME/`, while other output (JSON,
      LaTeX, source code for the MEX files) will be under `FILENAME/`.

    - The macro-generated output is now more readable (no more line numbers and
      empty lines). The old behaviour can be restored using the `linemacro`
      option (see above).

    - Ability to call the preprocessor by passing the `.mod` file as a string
      argument from the macOS or GNU/Linux command line.

 - dseries classes

    - New functionalities and efficiency improvements.

    - Complete rewrite using the new `classdef` syntax and exploiting in place
      modifications when possible.

    - Integration of the `dates` classes within `dseries`.

 - Reporting classes

    - Automatically create titlepage with page numbers/page titles.

    - Allow for the removal of headers and footers from a given page.

    - Allow user to set page number.

    - Split up report output. Create new files for the preamble, the body of
      the report, and each individual page of the report.

    - The classes have been converted to the new `classdef` syntax.

 - Misc

    - External functions can be located in MATLAB/Octave namespaces.

    - Improvements to the balanced growth path test that is performed after
      Dynare has detrended the model (given the trends on variables declared by
      the user): the default tolerance has been raised, and a different value
      can be set with new option `balanced_growth_test_tol` to the `model`
      block; as a consequence, failing the test is now an error again.

    - New collection of MATLAB/Octave utilities to retrieve and alter objects:
      `get_irf`, `get_mean`, `get_shock_stderr_by_name`, `get_smooth`,
      `get_update`, `set_shock_stderr_value`.

    - ⚠ Previously, when some MEX files were missing, Dynare would
      automatically fall back to slower M-file functional alternative; this is
      no longer the case. It is however still possible to manually add these
      alternatives in the MATLAB/Octave path (they are located under
      `matlab/missing/mex`; this only applies to the `mjdgges`, `gensylv`,
      `A_times_B_kronecker_C`, `sparse_hessian_times_B_kronecker_C` and
      `local_state_space_iteration_2` DLLs).


Since there are a few backward-incompatible changes in this release, users may
want to have a look at the [upgrade
guide](https://git.dynare.org/Dynare/dynare/-/wikis/BreakingFeaturesIn4.6) to
adapt their existing codes.


373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418
Bugs that were present in 4.5.7 and that are fixed in 4.6.0
-----------------------------------------------------------

* Estimation: the check for stochastic singularity erroneously would only take
  estimated measurement error into account.
* Estimation: if the Hessian at the mode was not positive definite, the Laplace
  approximation returned a complex number, but only displayed the real-valued
  part.
* Conditional Forecasting: using one period only would result in a crash.
* First-order approximation was not working with purely forward-looking models.
* The preprocessor would not allow for inline comments including macro
  statements.
* Using the `STEADY_STATE()` operator on exogenous variables would lead to
  crashes in stochastic simulations.
* `moment_calibration`: for autocorrelation functions, the x-axis labeling had
  the wrong order.
* `plot_identification`: placement of white dots indicating infinite values was
  incorrect
* Automatic detrending would sometime refuse to detrend model despite the user
  having given correct trends.
* Using `use_dll` + `fast` options would not always recompile the model when
  the equations were changed.
* Under certain circumstances, the combination of `bytecode` and
  `stack_solve_algo=1` options could lead to crashes or wrong results.


References
----------

 - Komunjer, I. and S. Ng (2011), “[Dynamic Identification of Dynamic
   Stochastic General Equilibrium
   Models](https://www.onlinelibrary.wiley.com/doi/abs/10.3982/ECTA8916),”
   *Econometrica*, 79(6), 1995–2032

 - Qu, Z. and D. Tkachenko (2012), “[Identification and frequency domain
   quasi‐maximum likelihood estimation of linearized dynamic stochastic
   general equilibrium
   models](https://onlinelibrary.wiley.com/doi/abs/10.3982/QE126),”
   *Quantitative Economics*, 3(1), 95–132

 - Mutschler, W. (2015), “[Identification of DSGE models—The effect of
   higher-order approximation and
   pruning](https://www.sciencedirect.com/science/article/pii/S0165188915000731),”
   *Journal of Economic Dynamics and Control*, 56, 34–54


Stéphane Adjemian's avatar
Stéphane Adjemian committed
419 420 421 422 423 424 425
Announcement for Dynare 4.5.7 (on 2019-02-06)
=============================================

We are pleased to announce the release of Dynare 4.5.7.

This is a bugfix release.

426
The Windows packages are already available for download at: <http://www.dynare.org/download/dynare-stable>.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

This release is compatible with MATLAB versions 7.5 (R2007b) to 9.4 (R2018b)
and with GNU Octave versions 4.4.1.

Here is a list of the problems identified in version 4.5.6 and that have been
fixed in version 4.5.7:

 - The mex-file conducting the QZ decomposition erroneously applied
   the `qz_criterium` to the square absolute value of eigenvalues
   instead of the absolute value itself (as done in mjdgges.m and the
   AIM solver).

 - In pathological cases, `mode_compute=5` (`newrat`) might enter an
   infinite loop.

 - `discretionary_policy` might erroneously state that the derivatives
   of the objective function are non-zero if there are NaN present.

 - Dynare++, when conducting the QZ decomposition, erroneously applied
   the `qz_criterium` to the square absolute value of eigenvalues
   instead of the absolute value itself.

 - Dynare++: IRFs were incorrectly computed.

 - `dynare_sensitivity` did not display the figures of
   `irf_calibration`, it only stored them on the disk.

 - Scatter plots generated by `dynare_sensitivity` did not correctly
   display LaTeX names.

 - Parameter updating via steady state files did not correctly work in
460
   case of using `[static]`/`[dynamic]` equation tags.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481

 - Memory leaks in `k_order_pert` (used by higher order stochastic
   simulations) could lead to crashes.

 - Predetermined variables were not properly set when used in model
   local variables.

 - Posterior moment computation did not correctly update the
   covariance matrix of exogenous shocks during posterior sampling.

 - Dynare was crashing with a cryptic message if a non estimated
   parameter was initialized in the `estimated_params_init` block.

 - The `forecast` command crashed if the model was declared as linear
   and contained deterministic exogenous variables.

 - Block decomposition is broken when used in conjunction with
   `varexo_det`.

 - The model was not correctly specified when `identification` was run
   without another stochastic command in the `.mod` file
482
   (*e.g.* `estimation`, `stoch_simul`, etc.).
Stéphane Adjemian's avatar
Stéphane Adjemian committed
483 484 485 486 487 488 489 490 491 492 493

 - Realtime annualized shock decompositions added the wrong steady state
   value.

 - `mh_recover` option crashed when using slice sampler.

 - x-axis values in plots of moment restrictions were wrong for
   autocovariances.



494 495 496 497 498 499 500 501
Announcement for Dynare 4.5.6 (on 2018-07-25)
=============================================

We are pleased to announce the release of Dynare 4.5.6.

This is a bugfix release.

The Windows packages are already available for download at:
502
<http://www.dynare.org/download/dynare-stable>.
503 504 505

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

506
This release is compatible with MATLAB versions 7.5 (R2007b) to 9.4 (R2018a)
507 508 509 510 511 512 513 514 515
and with GNU Octave versions 4.4.

Here is a list of the problems identified in version 4.5.5 and that have been
fixed in version 4.5.6:

 - TaRB sampler: incorrect last posterior was returned if the last draw was
   rejected.

 - Fixed online particle filter by drawing initial conditions in the prior
Stéphane Adjemian's avatar
Stéphane Adjemian committed
516
   distribution.
517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538

 - Fixed evaluation of the likelihood in non linear / particle filters.

 - Added missing documented `montecarlo` option in Gaussian Filter and
   Nonlinear Kalman Filter.

 - Added back a flag to deal with errors on Cholesky decomposition in the
   Conditional Particle Filter.

 - Macroprocessor `length()` operator was returning 1 when applied to a
   string. Macroprocessor now raises an error when `length()` operator is
   called on an integer and return the number of characters when applied to a
   string.

 - `mode_compute=8`: the error code during mode-finding was not correctly
   handled, resulting in crashes.

 - Identification was not correctly displaying a message for collinear parameters
   if there was no unidentified parameter present.



Stéphane Adjemian's avatar
Stéphane Adjemian committed
539 540 541 542 543 544 545 546
Announcement for Dynare 4.5.5 (on 2018-06-08)
=============================================

We are pleased to announce the release of Dynare 4.5.5.

This is a bugfix release.

The Windows packages are already available for download at:
547
<http://www.dynare.org/download/dynare-stable>.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
548 549 550

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

551
This release is compatible with MATLAB versions 7.5 (R2007b) to 9.4 (R2018a)
Stéphane Adjemian's avatar
Stéphane Adjemian committed
552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
and with GNU Octave versions 4.2.

Here is a list of the problems identified in version 4.5.4 and that have been
fixed in version 4.5.5:

 - Identification was crashing during prior sampling if `ar` was initially too
   low.

 - The `align` method on `dseries` did not return a functional second `dseries`
   output.

 - Predetermined variables were not properly set when used in model local
   variables.

 - `perfect_foresight_solver` with option `stack_solve_algo=7` was not working
   correctly when an exogenous variable has a lag greater than 1.

 - `identification` with `prior_mc` option would crash if the number of moments
   with non-zero derivative is smaller than the number of parameters.

 - Calling several times `normcdf` or `normpdf` with the same arguments in a
   model with block decomposition (but not bytecode) was leading to incorrect
   results.



578 579 580 581 582 583 584 585
Announcement for Dynare 4.5.4 (on 2018-01-29)
=============================================

We are pleased to announce the release of Dynare 4.5.4.

This is a bugfix release.

The Windows packages are already available for download at:
586
<http://www.dynare.org/download/dynare-stable>.
587 588 589

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

590
This release is compatible with MATLAB versions 7.5 (R2007b) to 9.3 (R2017b)
591 592 593 594 595 596 597 598 599 600 601 602 603 604 605
and with GNU Octave versions 4.2.

Here is a list of the problems identified in version 4.5.3 and that have been
fixed in version 4.5.4:

 - The `type` option of `plot_shock_decomposition` was always set to `qoq` regardless of what is specified.

 - Bug in GSA when no parameter was detected below pvalue threshold.

 - Various bug fixes in shock decompositions.

 - Bug in reading in macro arrays passed on `dynare` command line via the `-D` option.

 - Estimation with missing values was crashing if the `prefilter` option was used.

606
 - Added a workaround for a difference in behaviour between Octave and MATLAB regarding the creation
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
   of function handles for functions that do not exist in the path. With Octave 4.2.1, steady state
   files did not work if no auxiliary variables were created.

 - The `stoch_simul` command was crashing with a cryptic message if option `order=3` was used without
   setting `k_order_solver`.

 - In cases where the prior bounds are infinite and the mode is estimated at exactly 0, no `mode_check`
   graphs were displayed.

 - Parallel execution of MCMC was broken in models without auxiliary variables.

 - Reading data with column names from Excel might crash.

 - The multivariate Kalman smoother was crashing in case of missing data in the observations and
   `Finf` became singular.

 - The `plot_shock_decomposition` command ignored various user-defined options like `fig_name`,
   `use_shock_groups` or `interactive` and instead used the default options.

626
 - Nested `@#ifdef` and `@#ifndef` statements don’t work in the macroprocessor.
627 628 629



630 631 632 633 634 635 636 637 638
Announcement for Dynare 4.5.3 (on 2017-10-19)
=============================================

We are pleased to announce the release of Dynare 4.5.3.

This is a bugfix release. It comes less than 24 hours after the previous release,
because version 4.5.2 was affected by a critical bug for MATLAB older than R2016b.

The Windows packages are already available for download at:
639
<http://www.dynare.org/download/dynare-stable>.
640 641 642

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

643
This release is compatible with MATLAB versions 7.5 (R2007b) to 9.3 (R2017b)
644 645
and with GNU Octave versions 4.2.

Stéphane Adjemian's avatar
Stéphane Adjemian committed
646 647
Here is a list of the problems identified in version 4.5.2 and that have been
fixed in version 4.5.3:
648 649


650
 - `isfile` routine was failing with MATLAB older than R2016b. This bug did not
651 652 653 654
   affect Octave.



655 656 657 658 659 660 661 662
Announcement for Dynare 4.5.2 (on 2017-10-19)
=============================================

We are pleased to announce the release of Dynare 4.5.2.

This is a bugfix release.

The Windows packages are already available for download at:
663
<http://www.dynare.org/download/dynare-stable>.
664 665 666

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

667
This release is compatible with MATLAB versions 7.5 (R2007b) to 9.3 (R2017b)
668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714
and with GNU Octave versions 4.2.

Here is a list of the problems identified in version 4.5.1 and that have been
fixed in version 4.5.2:


 - Fixed bug in perfect foresight solver:

   + If expected shocks were declared after the terminal period, as specified
   by the `periods` option, Dynare was crashing.

   + Models declared with the `linear` option were crashing if exogenous
   variables were present with a lead or lag.

 - After ML or Bayesian estimation when the smoother option or `mh_replic=0`
   were not specified, not all smoothed measurement errors were displayed.

 - Fixed error in reference manual about the `conditional_forecasts` command.

 - Fixed smoother behaviour, provide informative error instead of crashing when
   model cannot be solved.

 - The `nopathchange` preprocessor option was always triggered, regardless of
   whether it was passed or not.

 - When `ramsey_policy` is used, allow state variables to be set in `histval`
   block.

 - `histval` erroneously accepted leads, leading to cryptic crashes.

 - The prior MC draws from previous runs were not deleted, potentially
   resulting in loading stale files.

 - `estim_params_` was being declared `global` more than once.

 - Fixed crashes happening when simulating linear models with order>1.

 - Make empirical moments independent of `simul_replic`, as stated in the
   reference manual, by outputting moments computed with the first simulated
   sample.

 - The `prior_function` required a preceding `estimation`-command to properly
   set up the prior.

 - If the mode for a parameter was at exactly 0, `mode_check` was crashing.

 - Fixed `get_posterior_parameters`-routine which should not do more than
Stéphane Adjemian's avatar
Stéphane Adjemian committed
715
   getting parameters. As a consequense, the `shock_decomposition`-command
716 717 718 719 720 721 722
   did not correctly set the `parameter_set` for use in subsequent function
   calls if shocks are correlated or measurement error is present.

 - Fixed bug in Ramsey problem with constraints both on a policy instrument and
   another variable. Note that the constraint on a variable that is not an
   instrument of the Ramsey problem must be written with an equation tag in the
   model block.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
723

724 725 726 727 728
 - Fixed bug in Ramsey problem with constraints on policy instrument.

 - Fixed crash with optimizer 5 when not used with DSGE model at order 1.

 - Fixed mex file used for third order approximation (was crashing on
729
   MATLAB/Windows 7).
730 731 732



733 734 735 736 737 738 739 740
Announcement for Dynare 4.5.1 (on 2017-08-24)
=============================================

We are pleased to announce the release of Dynare 4.5.1.

This is a bugfix release.

The Windows packages are already available for download at:
741
<http://www.dynare.org/download/dynare-stable>.
742 743 744

The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should follow soon.

745
This release is compatible with MATLAB versions 7.5 (R2007b) to 9.2 (R2017a)
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775
and with GNU Octave versions 4.2.

Here is a list of the problems identified in version 4.5.0 and that have been
fixed in version 4.5.1:


 - Fixed out of memory issue with simpsa optimization algorithm.

 - Added missing plots for measurement errors with `generate_trace_plot`
   command.

 - Posterior moments after MCMC for very big models were not correctly computed
   and their plotting might crash Dynare.

 - Results of the posterior conditional variance decomposition after MCMC were
   not correctly computed.

 - Options `use_shock_groups` and `colormap` of the `shock_decomposition`
   command were not working.

 - Added a clean error message if sensitivity toolbox is used with recursive
   estimation.

 - Computation of posterior filtered variables was crashing in models with only
   one variable.

 - Fixed various typos and errors in the reference manual.



776
Announcement for Dynare 4.5.0 (on 2017-06-11)
Stéphane Adjemian's avatar
Stéphane Adjemian committed
777 778 779 780 781 782 783
=============================================

We are pleased to announce the release of Dynare 4.5.0.

This major release adds new features and fixes various bugs.

The Windows packages are already available for download at:
784
<http://www.dynare.org/download/dynare-stable>.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
785 786 787 788 789

The Mac and Debian/Ubuntu packages should follow soon.

All users are strongly encouraged to upgrade.

790
This release is compatible with MATLAB versions ranging from 7.5 (R2007b) to
Stéphane Adjemian's avatar
Stéphane Adjemian committed
791 792 793 794 795 796 797
9.2 (R2017a) and with GNU Octave version 4.2.

Here is the list of major user-visible changes:


 - Ramsey policy

798
   + Added command `ramsey_model` that builds the expanded model with
799
     FOC conditions for the planner’s problem but doesn’t perform any
800 801
     computation. Usefull to compute Ramsey policy in a perfect
     foresight model,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
802

803 804
   + `ramsey_policy` accepts multipliers in its variable list and
     displays results for them.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
805 806 807 808


 - Perfect foresight models

809 810 811 812 813 814 815 816
   + New commands `perfect_foresight_setup` (for preparing the
     simulation) and `perfect_foresight_solver` (for computing it). The
     old `simul` command still exist and is now an alias for
     `perfect_foresight_setup` + `perfect_foresight_solver`. It is no
     longer possible to manipulate by hand the contents of
     `oo_.exo_simul` when using `simul`. People who want to do
     it must first call `perfect_foresight_setup`, then do the
     manipulations, then call `perfect_foresight_solver`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
817

818 819 820
   + By default, the perfect foresight solver will try a homotopy
     method if it fails to converge at the first try. The old behavior
     can be restored with the `no_homotopy` option,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
821

822 823
   + New option `stack_solve_algo=7` that allows specifying a
     `solve_algo` solver for solving the model,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
824

825 826
   + New option `solve_algo` that allows specifying a solver for
     solving the model when using `stack_solve_algo=7`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
827

828 829
   + New option `lmmcp` that solves the model via a Levenberg-Marquardt
     mixed complementarity problem (LMMCP) solver,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
830

831 832
   + New option `robust_lin_solve` that triggers the use of a robust
     linear solver for the default `solve_algo=4`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
833

834 835
   + New options `tolf` and `tolx` to control termination criteria of
     solvers,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
836

837
   + New option `endogenous_terminal_period` to `simul`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
838

839 840
   + Added the possibility to set the initial condition of the
     (stochastic) extended path simulations with the histval block.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
841 842 843 844


 - Optimal simple rules

845
   + Saves the optimal value of parameters to `oo_.osr.optim_params`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
846

847 848
   + New block `osr_params_bounds` allows specifying bounds for the
     estimated parameters,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
849

850 851
   + New option `opt_algo` allows selecting different optimizers while
     the new option `optim` allows specifying the optimizer options,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
852

853 854 855
   + The `osr` command now saves the names, bounds, and indices for the
     estimated parameters as well as the indices and weights of the
     variables entering the objective function into `M_.osr`.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
856 857 858 859


 - Forecasts and Smoothing

860 861
   + The smoother and forecasts take uncertainty about trends and means
     into account,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
862

863 864
   + Forecasts accounting for measurement error are now saved in fields
     of the form `HPDinf_ME` and `HPDsup_ME`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
865

866 867
   + New fields `oo_.Smoother.Trend` and `oo_.Smoother.Constant` that
     save the trend and constant parts of the smoothed variables,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
868

869 870
   + new field `oo_.Smoother.TrendCoeffs` that stores the trend
     coefficients.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
871

872 873
   + Rolling window forecasts allowed in `estimation` command by
     passing a vector to `first_obs`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
874

875 876
   + The `calib_smoother` command now accepts the `loglinear`,
     `prefilter`, `first_obs` and `filter_decomposition` options.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
877 878 879 880


 - Estimation

881 882 883
   + New options: `logdata`, `consider_all_endogenous`,
     `consider_only_observed`, `posterior_max_subsample_draws`,
     `mh_conf_sig`, `diffuse_kalman_tol`, `dirname`, `nodecomposition`
Stéphane Adjemian's avatar
Stéphane Adjemian committed
884

885
   + `load_mh_file` and `mh_recover` now try to load chain’s proposal density,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
886

887 888 889
   + New option `load_results_after_load_mh` that allows loading some
     posterior results from a previous run if no new MCMC draws are
     added,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
890

891 892 893
   + New option `posterior_nograph` that suppresses the generation of
     graphs associated with Bayesian IRFs, posterior smoothed objects,
     and posterior forecasts,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
894

895 896
   + Saves the posterior density at the mode in
     `oo_.posterior.optimization.log_density`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
897

898 899
   + The `filter_covariance` option now also works with posterior
     sampling like Metropolis-Hastings,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
900

901 902
   + New option `no_posterior_kernel_density` to suppress computation
     of kernel density of posterior objects,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
903

904 905
   + Recursive estimation and forecasting now provides the individual
     `oo_` structures for each sample in `oo_recursive_`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
906

907
   + The `trace_plot` command can now plot the posterior density,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
908

909 910
   + New command `generate_trace_plots` allows generating all trace
     plots for one chain,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
911

912 913 914
   + New commands `prior_function` and `posterior_function` that
     execute a user-defined function on parameter draws from the
     prior/posterior distribution,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
915

916 917
   + New option `huge_number` for replacement of infinite bounds with
     large number during `mode_compute`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
918

919 920 921 922 923 924
   + New option `posterior_sampling_method` allows selecting the new
     posterior sampling options:
     `tailored_random_block_metropolis_hastings` (Tailored randomized
     block (TaRB) Metropolis-Hastings), `slice` (Slice sampler),
     `independent_metropolis_hastings` (Independent
     Metropolis-Hastings),
Stéphane Adjemian's avatar
Stéphane Adjemian committed
925

926 927 928 929
   + New option `posterior_sampler_options` that allow controlling the
     options of the `posterior_sampling_method`, its `scale_file`-option
     pair allows loading the `_mh_scale.mat`-file storing the tuned
     scale factor from a previous run of `mode_compute=6`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
930

931 932
   + New option `raftery_lewis_diagnostics` that computes *Raftery and Lewis
     (1992)* convergence diagnostics,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
933

934
   + New option `fast_kalman_filter` that provides fast Kalman filter
935
     using Chandrasekhar recursions as described in *Ed Herbst (2015)*,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
936

937 938
   + The `dsge_var` option now saves results at the posterior mode into
     `oo_.dsge_var`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
939

940 941
   + New option `smoothed_state_uncertainty` to provide the uncertainty
     estimate for the smoothed state estimate from the Kalman smoother,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
942

943
   + New prior density: generalized Weibull distribution,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
944

945 946
   + Option `mh_recover` now allows continuing a crashed chain at the
     last save mh-file,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
947

948 949 950
   + New option `nonlinear_filter_initialization` for the
     `estimation` command. Controls the initial covariance matrix
     of the state variables in nonlinear filters.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
951

952 953 954
   + The `conditional_variance_decomposition` option now displays
     output and stores it as a LaTeX-table when the `TeX` option is
     invoked,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
955

956 957
   + The `use_calibration` to `estimated_params_init` now also works
     with ML,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
958

959
   + Improved initial estimation checks.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
960 961 962 963


 - Steady state

964 965 966
   + The default solver for finding the steady state is now a
     trust-region solver (can be triggered explicitly with option
     `solve_algo=4`),
Stéphane Adjemian's avatar
Stéphane Adjemian committed
967

968 969
   + New options `tolf` and `tolx` to control termination criteria of
     solver,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
970

971 972
   + The debugging mode now provides the termination values in steady
     state finding.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
973 974 975 976


 - Stochastic simulations

977
   + New options `nodecomposition`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
978

979 980
   + New option `bandpass_filter` to compute bandpass-filtered
     theoretical and simulated moments,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
981

982 983
   + New option `one_sided_hp_filter` to compute one-sided HP-filtered
     simulated moments,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
984

985 986
   + `stoch_simul` displays a simulated variance decomposition when
     simulated moments are requested,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
987

988 989
   + `stoch_simul` saves skewness and kurtosis into respective fields
     of `oo_` when simulated moments have been requested,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
990

991 992
   + `stoch_simul` saves the unconditional variance decomposition in
     `oo_.variance_decomposition`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
993

994 995
   + New option `dr_display_tol` that governs omission of small terms
     in display of decision rules,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
996

997 998
   + The `stoch_simul` command now prints the displayed tables as LaTeX
     code when the new `TeX` option is enabled,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
999

1000 1001
   + The `loglinear` option now works with lagged and leaded exogenous
     variables like news shocks,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1002

1003 1004
   + New option `spectral_density` that allows displaying the spectral
     density of (filtered) endogenous variables,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1005

1006 1007
   + New option `contemporaneous_correlation` that allows saving
     contemporaneous correlations in addition to the covariances.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1008 1009 1010 1011


 - Identification

1012
   + New options `diffuse_filter` and `prior_trunc`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1013

1014 1015
   + The `identification` command now supports correlations via
     simulated moments,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1016 1017 1018 1019


 - Sensitivity analysis

1020
   + New blocks `irf_calibration` and `moment_calibration`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1021

1022
   + Outputs LaTeX tables if the new `TeX` option is used,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1023

1024
   + New option `relative_irf` to `irf_calibration` block.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1025 1026 1027 1028


 - Conditional forecast

1029 1030
   + Command `conditional_forecast` now takes into account `histval`
     block if present.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1031 1032 1033 1034


 - Shock decomposition

1035 1036
   + New option `colormap` to `shocks_decomposition` for controlling
     the color map used in the shocks decomposition graphs,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1037

1038
   + `shocks_decomposition` now accepts the `nograph` option,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1039

1040 1041
   + New command `realtime_shock_decomposition` that for each period `T= [presample,...,nobs]`
     allows computing the:
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1042

1043
     * realtime historical shock decomposition `Y(t|T)`, *i.e.* without observing data in `[T+1,...,nobs]`
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1044

1045
     * forecast shock decomposition `Y(T+k|T)`
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1046

1047 1048 1049 1050 1051 1052 1053 1054 1055
     * realtime conditional shock decomposition `Y(T+k|T+k)-Y(T+k|T)`

   + New block `shock_groups` that allows grouping shocks for the
     `shock_decomposition` and `realtime_shock_decomposition` commands,

   + New command `plot_shock_decomposition` that allows plotting the
     results from `shock_decomposition` and
     `realtime_shock_decomposition` for different vintages and shock
     groupings.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1056 1057 1058 1059


 - Macroprocessor

1060
   + Can now pass a macro-variable to the `@#include` macro directive,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1061

1062 1063 1064
   + New preprocessor flag `-I`, macro directive `@#includepath`, and
     dynare config file block `[paths]` to pass a search path to the
     macroprocessor to be used for file inclusion via `@#include`.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1065 1066 1067 1068


 - Command line

1069
   + New option `onlyclearglobals` (do not clear JIT compiled functions
1070
     with recent versions of MATLAB),
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1071

1072 1073
   + New option `minimal_workspace` to use fewer variables in the
     current workspace,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1074

1075 1076 1077
   + New option `params_derivs_order` allows limiting the order of the
     derivatives with respect to the parameters that are calculated by
     the preprocessor,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1078

1079 1080
   + New command line option `mingw` to support the MinGW-w64 C/C++
     Compiler from TDM-GCC for `use_dll`.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1081 1082 1083 1084


 - dates/dseries/reporting classes

1085
   + New methods `abs`, `cumprod` and `chain`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1086

1087
   + New option `tableRowIndent` to `addTable`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1088

1089 1090
   + Reporting system revamped and made more efficient, dependency on
     matlab2tikz has been dropped.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1091 1092 1093 1094


 - Optimization algorithms

1095
   + `mode_compute=2` Uses the simulated annealing as described by
1096
     *Corana et al. (1987)*,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1097

1098 1099
   + `mode_compute=101` Uses SOLVEOPT as described by *Kuntsevich and
     Kappel (1997)*,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1100

1101
   + `mode_compute=102` Uses `simulannealbnd` from MATLAB’s Global
1102
     Optimization Toolbox (if available),
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1103

1104 1105
   + New option `silent_optimizer` to shut off output from mode
     computing/optimization,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1106

1107 1108
   + New options `verbosity` and `SaveFiles` to control output and
     saving of files during mode computing/optimization.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1109 1110 1111 1112


 - LaTeX output

1113
   + New command `write_latex_original_model`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1114

1115 1116 1117
   + New option `write_equation_tags` to `write_latex_dynamic_model`
     that allows printing the specified equation tags to the generate
     LaTeX code,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1118

1119 1120
   + New command `write_latex_parameter_table` that writes the names and
     values of model parameters to a LaTeX table,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1121

1122 1123
   + New command `write_latex_prior_table` that writes the descriptive
     statistics about the prior distribution to a LaTeX table,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1124

1125 1126
   + New command `collect_latex_files` that creates one compilable LaTeX
     file containing all TeX-output.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1127 1128 1129 1130


 - Misc.

1131
   + Provides 64bit preprocessor,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1132

1133 1134
   + Introduces new path management to avoid conflicts with other
     toolboxes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1135

1136
   + Full compatibility with MATLAB 2014b’s new graphic interface,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1137

1138 1139
   + When using `model(linear)`, Dynare automatically checks
     whether the model is truly linear,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1140

1141 1142
   + `usedll`, the `msvc` option now supports `normcdf`, `acosh`,
     `asinh`, and `atanh`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1143

1144 1145 1146
   + New parallel option `NumberOfThreadsPerJob` for Windows nodes that
     sets the number of threads assigned to each remote MATLAB/Octave
     run,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1147

1148 1149
   + Improved numerical performance of
     `schur_statespace_transformation` for very large models,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1150

1151
   + The `all_values_required` option now also works with `histval`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1152

1153
   + Add missing `horizon` option to `ms_forecast`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1154

1155 1156 1157 1158
   + BVAR now saves the marginal data density in
     `oo_.bvar.log_marginal_data_density` and stores prior and
     posterior information in `oo_.bvar.prior` and
     `oo_.bvar.posterior`.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1159 1160 1161



1162
Bugs and problems identified in version 4.4.3 and that have been fixed in version 4.5.0:
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1163 1164 1165 1166


 - BVAR models

1167 1168
   + `bvar_irf` could display IRFs in an unreadable way when they moved from
     negative to positive values,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1169

1170 1171
   + In contrast to what is stated in the documentation, the confidence interval
     size `conf_sig` was 0.6 by default instead of 0.9.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1172 1173 1174 1175


 - Conditional forecasts

1176 1177 1178
   + The `conditional_forecast` command produced wrong results in calibrated
     models when used at initial values outside of the steady state (given with
     `initval`),
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1179

1180 1181
   + The `plot_conditional_forecast` option could produce unreadable figures if
     the areas overlap,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1182

1183
   + The `conditional_forecast` command after MLE crashed,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1184

1185 1186
   + In contrast to what is stated in the manual, the confidence interval size
     `conf_sig` was 0.6 by default instead of 0.8.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1187

1188 1189 1190
   + Conditional forecasts were wrong when the declaration of endogenous
     variables was not preceeding the declaration of the exogenous
     variables and parameters.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1191 1192 1193 1194


 - Discretionary policy

1195 1196
   + Dynare allowed running models where the number of instruments did not match
     the number of omitted equations,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1197

1198
   + Dynare could crash in some cases when trying to display the solution,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1199

1200 1201
   + Parameter dependence embedded via a `steady_state` was not taken into
     account, typically resulting in crashes.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1202 1203 1204

 - dseries class

1205 1206
   + When subtracting a dseries object from a number, the number was instead
     subtracted from the dseries object.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1207 1208 1209 1210


 - DSGE-VAR models

1211 1212
   + Dynare crashed when estimation encountered non-finite values in the Jacobian
     at the steady state,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1213

1214 1215 1216 1217
   + The presence of a constant was not considered for degrees of freedom
     computation of the Gamma function used during the posterior computation; due
     to only affecting the constant term, results should be be unaffected, except
     for model_comparison when comparing models with and without.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1218 1219 1220 1221


 - Estimation command

1222 1223
   + In contrast to what was stated in the manual, the confidence interval size
     `conf_sig` for `forecast` without MCMC was 0.6 by default instead of 0.9,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1224

1225
   + Calling estimation after identification could lead to crashes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1226

1227 1228 1229 1230
   + When using recursive estimation/forecasting and setting some elements of
     `nobs` to be larger than the number of observations T in the data,
     `oo_recursive_` contained additional cell entries that simply repeated the
     results obtained for `oo_recursive_T`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1231

1232 1233
   + Computation of Bayesian smoother could crash for larger models when
     requesting `forecast` or `filtered_variables`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1234

1235 1236
   + Geweke convergence diagnostics were not computed on the full MCMC chain when
     the `load_mh_file` option was used,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1237

1238 1239
   + The Geweke convergence diagnostics always used the default `taper_steps` and
   `geweke_interval`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1240

1241 1242
   + Bayesian IRFs (`bayesian_irfs` option) could be displayed in an unreadable
     way when they move from negative to positive values,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1243

1244 1245
   + If `bayesian_irfs` was requested when `mh_replic` was too low to compute
     HPDIs, plotting was crashing,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1246

1247 1248 1249
   + The x-axis value in `oo_.prior_density` for the standard deviation and
     correlation of measurement errors was written into a field
     `mearsurement_errors_*` instead of `measurement_errors_*`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1250

1251
   + Using a user-defined `mode_compute` crashed estimation,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1252

1253
   + Option `mode_compute=10` did not work with infinite prior bounds,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1254

1255 1256 1257
   + The posterior variances and covariances computed by `moments_varendo` were
     wrong for very large models due to a matrix erroneously being filled up with
     zeros,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1258

1259 1260
   + Using the `forecast` option with `loglinear` erroneously added the unlogged
     steady state,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1261

1262 1263
   + When using the `loglinear` option the check for the presence of a constant
     was erroneously based on the unlogged steady state,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1264

1265 1266
   + Estimation of `observation_trends` was broken as the trends specified as a
     function of deep parameters were not correctly updated during estimation,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1267

1268 1269 1270
   + When using `analytic_derivation`, the parameter values were not set before
     testing whether the steady state file changes parameter values, leading to
     subsequent crashes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1271

1272 1273 1274
   + If the steady state of an initial parameterization did not solve, the
     observation equation could erroneously feature no constant when the
     `use_calibration` option was used,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1275

1276 1277
   + When computing posterior moments, Dynare falsely displayed that moment
     computations are skipped, although the computation was performed correctly,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1278

1279 1280 1281
   + If `conditional_variance_decomposition` was requested, although all
     variables contain unit roots, Dynare crashed instead of providing an error
     message,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1282

1283 1284 1285
   + Computation of the posterior parameter distribution was erroneously based
     on more draws than specified (there was one additional draw for every Markov
     chain),
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1286

1287
   + The estimation option `lyapunov=fixed_point` was broken,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1288

1289
   + Computation of `filtered_vars` with only one requested step crashed Dynare,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1290

1291
   + Option `kalman_algo=3` was broken with non-diagonal measurement error,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1292

1293
   + When using the diffuse Kalman filter with missing observations, an additive
1294
     factor log(2π) was missing in the last iteration step,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1295

1296 1297
   + Passing of the `MaxFunEvals` and `InitialSimplexSize` options to
     `mode_compute=8` was broken,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1298

1299 1300
   + Bayesian forecasts contained initial conditions and had the wrong length in
     both plots and stored variables,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1301

1302 1303 1304
   + Filtered variables obtained with `mh_replic=0`, ML, or
     `calibrated_smoother` were padded with zeros at the beginning and end and
     had the wrong length in stored variables,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1305

1306
   + Computation of smoothed measurement errors in Bayesian estimation was broken,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1307

1308 1309 1310
   + The `selected_variables_only` option (`mh_replic=0`, ML, or
     `calibrated_smoother`) returned wrong results for smoothed, updated, and
     filtered variables,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1311

1312 1313
   + Combining the `selected_variables_only` option with forecasts obtained
     using `mh_replic=0`, ML, or `calibrated_smoother` leaded to crashes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1314

1315
   + `oo_.UpdatedVariables` was only filled when the `filtered_vars` option was specified,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1316

1317 1318 1319
   + When using Bayesian estimation with `filtered_vars`, but without
     `smoother`, then `oo_.FilteredVariables` erroneously also contained filtered
     variables at the posterior mean as with `mh_replic=0`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1320

1321 1322
   + Running an MCMC a second time in the same folder with a different number of
     iterations could result in crashes due to the loading of stale files,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1323

1324 1325 1326 1327 1328 1329
   + Results displayed after Bayesian estimation when not specifying
     the `smoother` option were based on the parameters at the mode
     from mode finding instead of the mean parameters from the
     posterior draws. This affected the smoother results displayed, but
     also calls to subsequent command relying on the parameters stored
     in `M_.params` like `stoch_simul`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1330

1331 1332 1333
   + The content of `oo_.posterior_std` after Bayesian estimation was based on
     the standard deviation at the posterior mode, not the one from the MCMC, this
     was not consistent with the reference manual,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1334

1335
   + When the initialization of an MCMC run failed, the metropolis.log file was
1336
     locked, requiring a restart of MATLAB to restart estimation,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1337

1338 1339
   + If the posterior mode was right at the corner of the prior bounds, the
     initialization of the MCMC erroneously crashed,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1340

1341
   + If the number of dropped draws via `mh_drop` coincided with the number of
1342
     draws in a `_mh`-file, `oo_.posterior.metropolis.mean` and
1343
     `oo_.posterior.metropolis.Variance` were NaN.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1344 1345 1346 1347


 - Estimation and calibrated smoother

1348 1349
   + When using `observation_trends` with the `prefilter` option, the mean shift
     due to the trend was not accounted for,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1350

1351 1352 1353
   + When using `first_obs`>1, the higher trend starting point of
     `observation_trends` was not taken into account, leading, among other things,
     to problems in recursive forecasting,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1354

1355 1356
   + The diffuse Kalman smoother was crashing if the forecast error variance
     matrix becomes singular,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1357

1358 1359
   + The multivariate Kalman smoother provided incorrect state estimates when
     all data for one observation are missing,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1360

1361 1362
   + The multivariate diffuse Kalman smoother provided incorrect state estimates
     when the `Finf` matrix becomes singular,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1363

1364 1365
   + The univariate diffuse Kalman filter was crashing if the initial covariance
     matrix of the nonstationary state vector is singular,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1366 1367 1368 1369


 - Forecats

1370 1371
   + In contrast to what is stated in the manual, the confidence interval size
     `conf_sig` was 0.6 by default instead of 0.9.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1372

1373 1374
   + Forecasting with exogenous deterministic variables provided wrong decision
     rules, yielding wrong forecasts.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1375

1376 1377
   + Forecasting with exogenous deterministic variables crashed when the
     `periods` option was not explicitly specified,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1378

1379 1380 1381
   + Option `forecast` when used with `initval` was using the initial values in
     the `initval` block and not the steady state computed from these initial
     values as the starting point of forecasts.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1382 1383 1384 1385


 - Global Sensitivity Analysis

1386
   + Sensitivity with ML estimation could result in crashes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1387

1388
   + Option `mc` must be forced if `neighborhood_width` is used,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1389

1390
   + Fixed dimension of `stock_logpo` and `stock_ys`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1391

1392
   + Incomplete variable initialization could lead to crashes with `prior_range=1`.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1393 1394 1395 1396


 - Indentification

1397 1398 1399
   + Identification did not correctly pass the `lik_init` option,
     requiring the manual setting of `options_.diffuse_filter=1` in
     case of unit roots,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1400

1401 1402
   + Testing identification of standard deviations as the only
     parameters to be estimated with ML leaded to crashes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1403

1404 1405 1406
   + Automatic increase of the lag number for autocovariances when the
     number of parameters is bigger than the number of non-zero moments
     was broken,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1407

1408
   + When using ML, the asymptotic Hessian was not computed,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1409

1410 1411
   + Checking for singular values when the eigenvectors contained only
     one column did not work correctly,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1412 1413 1414 1415


 - Model comparison

1416
   + Selection of the `modifiedharmonicmean` estimator was broken,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1417 1418 1419 1420


 - Optimal Simple Rules

1421 1422 1423
   + When covariances were specified, variables that only entered with
     their variance and no covariance term obtained a wrong weight,
     resulting in wrong results,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1424

1425 1426 1427 1428
   + Results reported for stochastic simulations after `osr` were based
     on the last parameter vector encountered during optimization,
     which does not necessarily coincide with the optimal parameter
     vector,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1429

1430
   + Using only one (co)variance in the objective function resulted in crashes,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1431

1432
   + For models with non-stationary variables the objective function was computed wrongly.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1433 1434 1435 1436


 - Ramsey policy

1437 1438
   + If a Lagrange multiplier appeared in the model with a lead or a lag
     of more than one period, the steady state could be wrong.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1439

1440 1441
   + When using an external steady state file, incorrect steady states
     could be accepted,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1442

1443 1444
   + When using an external steady state file with more than one
     instrument, Dynare crashed,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1445

1446 1447
   + When using an external steady state file and running `stoch_simul`
     after `ramsey_planner`, an incorrect steady state was used,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1448

1449 1450
   + When the number of instruments was not equal to the number of
     omitted equations, Dynare crashed with a cryptic message,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1451

1452
   + The `planner_objective` accepted `varexo`, but ignored them for computations,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1453 1454 1455 1456


 - Shock decomposition

1457 1458
   + Did not work with the `parameter_set=calibration` option if an
     `estimated_params` block is present,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1459

1460
   + Crashed after MLE.
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1461 1462 1463 1464


 - Perfect foresight models

1465 1466
   + The perfect foresight solver could accept a complex solution
     instead of continuing to look for a real-valued one,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1467

1468
   + The `initval_file` command only accepted column and not row vectors,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1469

1470
   + The `initval_file` command did not work with Excel files,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1471

1472 1473 1474
   + Deterministic simulations with one boundary condition crashed in
     `solve_one_boundary` due to a missing underscore when passing
     `options_.simul.maxit`,
Stéphane Adjemian's avatar
Stéphane Adjemian committed
1475

1476 1477
   + Deterministic simulation with exogenous variables lagged by more
     than one period crashed,