# This page documents known bugs in Dynare stable.
* The `keep_kalman_algo_if_singularity_is_detected` option of the `estimation` command did not work as advertised (option will be removed in Dynare 7, see 41a3bbd7)
* The `discretionary_policy` command did not accept the `planner_discount_latex_name` option, contrary to what the documentation says (bug fixed in preprocessor@f7917e9bd4369f25a20ce81e617014fdfe4f5718)
* The `outfile` option of the `smoother2histval` command would produce an incorrect file on models with two lags or more, leading to incorrect results when loading that file with `histval_file` (bug fixed in 618712ad9fb9a128545ae9ee8ca01ca2900a3af9)
* The point forecasts triggered by `forecast` option of `estimation` did not take shock uncertainty into account (bug fixed in f03929f3)
* The `nodecomposition` option of `estimation` did not suppress all variance decompositions (bug fixed in 6d2828287bf04f7e85369d99fcc69e411f24ad91)
* The `example1_reporting.mod` file was broken (bug fixed in b3e7d59b)
* The `EndTemperature` option of `mode_compute=10` (`simpsa`) was broken (bug fixed in d802ed7c)
* Optimization would, in rare cases, erroneously apply an infinite penalty function value (bug fixed in 46809c41)
* The `mh_posterior_mode_estimation` was broken in conjunction with `method_of_moments` (bug fixed in e3c0572b)
* The following problems with non-linear filters were fixed:
*`filter_algorithm=cpf` with `proposal_approximation=montecarlo` returned incorrect results (bug fixed in 72b00500)
* The `particle_filter_options``unscented_beta` and `unscented_kappa` did not accept values of 0 (bug fixed in daaf9def)
*`filter_algorithm=gf` and `filter_algorithm=gmf` did not correctly handle correlated measurement error (bug fixed in 62fe5425)
* The `estimation` command did not allow the `pruning` option to be set for higher-order estimation (bug fixed in 4601ee53)
* The manual incorrectly stated that `1,000` instead of `5,000` particles was the default (bug fixed in ea18e1ec)
* OccBin would occasionally crash upon encountering invalid parameter vectors (bug fixed in 88f1b6b5)
* In rare cases, the multivariate diffuse Kalman filter/smoother (`kalman_algo=3`) did not correctly exit the diffuse stage (bug fixed in 32c421a3)
*`forecast` with `varexo_det`
* without measurement errors stored the upper bound of the confidence interval in the `HPDinf` field and the lower bound in the `HPDsup` field (bug fixed in 20c38954)
* with multiple measurement errors would crash if not all observables were requested as outputs (bug fixed in dd94003e)
* would assign the output to the wrong variable names if a variable list was specified after the command (bug fixed in 89ff657e)
* The confidence bands of classical forecasts with and without `varexo_det` had incorrect coverage (bug fixed in f3c08902)
* The `posterior_sampling_method='slice'` did not save all files in the correct folder (bug fixed in 9801ad62)
* The `mh_recover` option did not work correctly for `posterior_sampling_method='slice'` when `save_tmp_file` was used (bug not yet fixed, see https://git.dynare.org/Dynare/dynare/-/merge_requests/2441)
*`dynare_solve`, used in e.g. numerical steady state computations,
* only directly accepted correct initial values if the Jacobian contained `NaN` or `Inf`. Otherwise, the numerical solver was started (bug fixed in 2b5ec0ba).
* did not honor the `solve_randomize_initial_guess` option if no Jacobian was requested (bug fixed in 501556f5)
* erroneously accepted initial guesses with complex values when randomizing the starting point (bug fixed in b7f5ac8d)
* erroneously accepted initial guesses that contained `Inf` (bug fixed in 45c813df)
* Using the `forecast` command with a model solved at `order=2` with `varexo_det`, Dynare would return uncertainty bands based on a first-order approximation (bug fixed for Dynare 7, see #1940)
* The results of mode-finding were not correctly stored in `oo_.posterior_mode` for shock and measurement error correlations (bug fixed in 7a8d8539)
* OccBin would not correctly sum the likelihood of detected multiple solution paths (bug fixed in c6e143e0 for Dynare 7)
* Using the `forecast` command with a model solved at `order=2` with
`varexo_det`, Dynare would return uncertainty bands based on a first-order
approximation (bug will be fixed in Dynare 7, see #1940)
* OccBin would not correctly sum the likelihood of detected multiple solution
paths (bug fixed in c6e143e0 for Dynare 7)
### Low priority bugs that won't be fixed soon:
* Forecasts are only based on the DSGE-model, not the DSGE-VAR (bug not yet fixed; see #819)
* The `contemp_reduced_form` option of the `sbvar` command is broken (bug not yet fixed, see #1716)
