Stéphane Adjemian (Charybdis)
9fbef0c107
Removed penalty_hessian routine.
...
+ Code factorization.
+ Added an option for using the penalized objective when computing numerically
the hessian at the mode.
Previous behaviour (introduced with penalty_hessian routine) was to compute the
hessian matrix at the mode with the penalized objective function (instead of
the original objective function). This behaviour hides problematic situations,
where the computed hessian (using the original objective) would not be full
rank. For instance, if the estimation ends up with a parameter on (or very
close to) the bounds of its possible values (which is often not a desirable
outcome), the estimated posterior variance would be zero for this
parameter (with the original objective) because the hessian is not finite in
this direction, while the posterior variance would be positive if the penalized
objective is used instead. But this estimate would not be reliable by
construction of the penalty which is quite ad-hoc (more fundamentally I do not
think that there exists any rational for approximating the covariance matrix
with the inverse of the hessian matrix if the mode is on the border of the set
of possible values).
This commit restore the behaviour previous to 2446ab02ba
.
An option is available for computing the hessian with the penalized
objective function.
2017-02-08 13:07:25 +00:00
Johannes Pfeifer
452d3cd986
Implement preprocessor interface for posterior_nograph option
2016-12-26 13:07:46 +01:00
Johannes Pfeifer
352786c2b4
load_mh_file: recompute results by default and load on request
...
Prevents general overwriting of results in oo_ and having stale results
2016-12-19 19:37:34 +01:00
Johannes Pfeifer
5515da191a
Add Raftery/Lewis (1992) convergence diagnostics
2016-12-17 17:22:05 +01:00
Johannes Pfeifer
d5e20752ac
robust_lin_solve: Fix mistakes introduced in 4b83c1bf76
...
Interface was still missing and documentation of option was in wrong place in manual and wrong
2016-12-01 17:31:29 +01:00
Johannes Pfeifer
98cf1bfc1d
Add smoothed state uncertainty to Kalman smoother routines
2016-11-04 09:23:55 +01:00
Houtan Bastani
4b3c98d953
preprocessor: add keep_kalman_algo_if_singularity_is_detected option to estimation. #1294
2016-09-30 14:31:13 +02:00
Houtan Bastani
ed549d3ba7
preprocessor: add tolx to perfect_foresight_solver
2016-07-21 09:40:19 -04:00
Johannes Pfeifer
1c420e6056
Allow for parameter_set=mle_mode in shock_decomposition
...
Required after writing results into oo_.mle_mode for ML estimation
2016-06-17 00:21:08 +02:00
Stéphane Adjemian (Charybdis)
15f83d0e18
Added new option for setting weights in CPF algorithm.
2016-06-01 16:29:53 +02:00
Houtan Bastani
8b85ca19bf
preprocessor: support sub lists in posterior_sampling_opt option, hence removing tarb_optim. #1177
2016-05-19 14:37:05 +02:00
Houtan Bastani
cf6e3cb14c
preprocessor: rework posterior sampler options, #1177
2016-05-19 14:37:05 +02:00
Houtan Bastani
42895ce32c
Revert "preprocessor: add params_derivs_order to identification and sensitivity. #1187"
...
This reverts commit 6192a6a776
.
2016-05-18 10:33:45 +02:00
Houtan Bastani
6192a6a776
preprocessor: add params_derivs_order to identification and sensitivity. #1187
2016-05-17 17:18:16 +02:00
Houtan Bastani
626f8af374
preprocessor: add analytic_derivation_mode option to identification and sensitivity. #1187
2016-05-17 15:32:51 +02:00
Houtan Bastani
a4b6dfa3f3
preprocessor: interface for osr_params_bounds. #948
2016-05-10 18:01:00 +02:00
Michel Juillard
800d79d6e5
add option colormap for shocks_decomposition
2016-04-15 09:02:53 +02:00
Michel Juillard
de567377c3
adding groups of shocks for shock_decomposition
...
consistency checks, documentation and test case
are still missing
2016-04-15 09:02:53 +02:00
Sébastien Villemot
2643bfaa40
Fix compatibility with Flex 2.6.
...
In Flex 2.6, the C++ scanner changed. yyout is now a reference (and no
longer a pointer).
2016-03-14 16:28:24 +01:00
Houtan Bastani
c9ca46f228
preprocessor: allow arbitrary partitioning of variables (removes long_name keyword)
2016-02-26 16:47:17 +01:00
Michel Juillard
c373d1e1be
adding new option 'fast_kalman_filter' implementing Ed Herbst 2012 approach
2015-11-28 17:38:00 +01:00
Houtan Bastani
6ef4371630
Revert "preprocessor: add utf-8 tokenizing"
...
This reverts commit f2cc9d3899
.
This commit causes warnings in Flex
2015-10-30 18:27:41 +01:00
Houtan Bastani
fd5ce1366e
preprocessor: change prior_posterior_function into two arguments. closes #1076
2015-10-14 11:02:35 +02:00
Houtan Bastani
145e2d5542
preprocessor: prior_posterior_function: change option 'prior_posterior_sampling_draws' to 'sampling_draws' #1076
2015-10-13 17:41:31 +02:00
Houtan Bastani
b4aa21f018
preprocessor: interface for prior_posterior_function command. closes #1076
2015-10-13 17:16:10 +02:00
Johannes Pfeifer
b7cbb563d6
Allow suppressing density of smoother and forecast objects
2015-10-12 15:34:52 +02:00
Houtan Bastani
8d54aeeaf6
preprocessor: add relative_irf option to irf_calibration block. closes #720
2015-09-07 11:49:17 +02:00
Houtan Bastani
11eaf27453
Merge branch 'julia'
2015-09-01 11:35:16 +02:00
Houtan Bastani
154980ec8c
preprocessor: posterior_kernel_density option to estimation. #1035
2015-08-25 11:09:08 +02:00
Michel Juillard
511801c903
finalize svar_global_identification_check
2015-08-24 14:53:27 +02:00
Houtan Bastani
f2cc9d3899
preprocessor: add utf-8 tokenizing
2015-08-17 17:09:40 +02:00
Johannes Pfeifer
698a44c98a
Add option for storing contemporaneous correlation
2015-08-11 11:27:55 +02:00
Houtan Bastani
e13a59eb64
preprocessor: add one_sided_hp_filter to stoch_simul. #1011
2015-08-03 17:34:34 +02:00
Houtan Bastani
d9a1b99899
preprocessor: add bandpass_filter option to stoch_simul. #1011
2015-08-03 17:28:55 +02:00
Houtan Bastani
d5cf474b6b
preprocessor: add spectral_density option to stoch_simul. #254
2015-08-03 13:02:17 +02:00
Stéphane Adjemian (Charybdis)
8007f508a2
New option linear_approximation for perfect foresight models.
...
This approach only requires one evaluation of the dynamic model (and its
jacobian) instead of T (the number of perdiods). Also (because the model
is linear) the equilibrium paths are obtained by inverting the jacobian
of the stacked equations (no need for a Newton algorithm).
Only available with stack_solve_algo==0 (which is the default algorithm
for solving perfect foresight models).
If possible, the option is triggered automatically if the model is
declared linear.
TODO:
* Write a linear version of perfect_foresight_problem routine.
* Evaluate the approxilation error (just need to evaluate the system of
stacked non linear equations).
2015-07-07 17:55:41 +02:00
Stéphane Adjemian
71700dff76
Merge pull request #942 from JohannesPfeifer/TaRB_integration
...
Integrate the TaRB-algorithm into Dynare
2015-06-19 14:30:56 +02:00
Houtan Bastani
aeb1542b48
preprocessor: add VerbatimStatement class. Closes #953
2015-06-16 12:48:32 +02:00
Johannes Pfeifer
55d44f0983
Add preprocessor option silent_optimizer
2015-06-08 16:48:57 +02:00
Michel Juillard
2f0cb39e14
adding options LMMCP and OCCBIN to perfect_foresight_solver and
...
extended_path
2015-06-02 17:38:22 +02:00
Michel Juillard
c22d31330c
Adding 'ramsey_constraints' block to declare contraints for Ramsey
...
problems. Note that LMMCP (solve_algo=10) still doesn't work in all cases.
2015-05-31 12:18:20 +02:00
Houtan Bastani
89828e56cc
preprocessor: add proprosal_distribution and student_degrees_of_freedom options to estimation. closes #925
2015-05-19 12:31:22 +02:00
Houtan Bastani
44790abbb7
preprocessor: add tarb_optim option to estimation. #940
2015-05-13 16:50:07 +02:00
Houtan Bastani
12be4e6672
preprocessor: add tarb_new_block_probability option to estimation. #940
2015-05-13 15:49:41 +02:00
Houtan Bastani
0d76e30248
preprocessor: add tarb_mode_compute option to estimation. #940
2015-05-13 15:46:52 +02:00
Houtan Bastani
e5df26a6e8
preprocessor: add use_tarb option to estimation. #940
2015-05-13 15:32:24 +02:00
Houtan Bastani
f239c1c368
preprocessor: add huge_number to estimation and osr. #924
2015-05-13 15:14:34 +02:00
Houtan Bastani
027f1302d1
preprocessor: add dr_display_tol to stoch_simul. #924
2015-05-13 15:12:39 +02:00
Houtan Bastani
18e9521450
preprocessor: add dirname option to estimation. closes #910
2015-04-27 11:59:21 +02:00
Stéphane Adjemian (Charybdis)
090c4fedbd
Added new option (diffuse_kalman_tol) and fixed tolerance paremeters in diffuse smoother routines.
2015-04-03 18:02:03 +02:00