Stéphane Adjemian (Charybdis)
1970951d1f
Merge branch 'johannes_fix_simulated_annealing_and_other_optimization_related_issues'
2015-04-16 22:31:07 +02:00
Stéphane Adjemian (Charybdis)
fff07396d2
Divided tests/optimizers/fs2000.mod into 12 mod files...
...
Otherwise the execution of the testsuite is not parallelized. Also, added a test for
mode_compute=1 (if fmincon from Mathworks' Optimization toolbox is
available).
2015-04-16 22:26:24 +02:00
Marco Ratto
b0f4f2af78
Harmonize use of front-end mcf_analysis.m in GSA and identification routines.
2015-04-16 18:25:35 +02:00
Marco Ratto
57e9b9fa8b
Revived reduced form mapping routine:
...
1) use new mcf_analysis.m;
2) use generic filtering when threshold does not work;
3) make log-tranformed mapping meaningful and comparable to baseline;
3) harmonize names of files saved and reduced number of directories created;
2015-04-16 16:13:16 +02:00
Marco Ratto
0faf6e1981
Allow inverse transform and set minimum threshold for lam
2015-04-16 16:13:00 +02:00
Marco Ratto
ae6302c348
Added output argument to mcf_analysis.m
2015-04-16 16:09:39 +02:00
Stéphane Adjemian
1509e07728
Merge pull request #884 from rattoma/gsa
...
fix graphic issues with eps
2015-04-16 12:44:03 +02:00
Marco Ratto
4e1e7732fd
fix graphic issues with eps
2015-04-16 12:30:27 +02:00
Stéphane Adjemian
fe243e5f8b
Merge pull request #883 from rattoma/gsa
...
missing bits in fixes to graphical issues #881 .
2015-04-16 09:16:50 +02:00
Marco Ratto
8d679bff6f
missing bits in fixes to graphical issues #881 .
2015-04-16 08:52:42 +02:00
Stéphane Adjemian (Charybdis)
31dc863077
Fixed graphic issue in tests/gsa/ls2003a.mod.
...
We still need to find out why eps format does not work under matlab...
2015-04-16 00:30:52 +02:00
Stéphane Adjemian
00c317e109
Merge pull request #882 from rattoma/gsa
...
fix graph output issues #881
2015-04-15 23:26:19 +02:00
Marco Ratto
41cebf7e4e
- fix nograph option in GSA graphical output
...
- fix graphical output issues on unix
2015-04-15 22:54:36 +02:00
Stéphane Adjemian (Charybdis)
79d0310c10
Added traps if some optimization algorithm are not availabe (testsuite).
2015-04-15 12:48:03 +02:00
Stéphane Adjemian (Charybdis)
f0770186ca
Exhanged mode_compute=2 and mode_compute=102.
...
These options are for the simulated annealing algorithm, the Mathworks' global optimization
toolbox is mandatory for mode_compute=102. The first option relies on a
free implementation (works also with Octave).
2015-04-15 12:42:08 +02:00
Stéphane Adjemian (Charybdis)
a7f0366980
Removed gsa/ls2003a.mod from the test suite (which causes a segmentation fault and breaks the test suite).
2015-04-15 12:31:09 +02:00
Stéphane Adjemian (Charybdis)
da2c529ee7
Merge branch 'marco_gsa_fixes'
2015-04-14 15:54:59 +02:00
Stéphane Adjemian (Charybdis)
58641d0efc
Added gsa/ls2003a.mod in testuite.
2015-04-14 15:51:50 +02:00
Stéphane Adjemian (Charybdis)
8022a181c8
Removed excel data file used in ls2003_bayesian_xls.mod.
...
The excel data file available here:
http://www.dynare.org/Datasets/data_ca1_xls.xlsx
is downloaded by Matlab/Octave when performing the test. The file is
deleted after the estimation command.
2015-04-14 15:12:25 +02:00
Johannes Pfeifer
62f81da6dd
Add unit test for recursive estimation with Excel data
2015-04-14 15:12:25 +02:00
Johannes Pfeifer
1a6afc0fdb
Restore compatibility of recursive estimation with new data interface
...
Use dataset_ structure instead of reconstructing data to plot. Eliminates crashes
2015-04-14 15:12:25 +02:00
Houtan Bastani
f5cf7617a6
preprocessor: method out of order
2015-04-14 14:47:26 +02:00
Marco Ratto
8908cfe0f4
Updated testing routine
2015-04-14 12:38:18 +02:00
Marco Ratto
0bd8fef202
- Various bug fixes on moment calibration
...
- Bug fixes on plots showing restrictions
- Improved and increased graphical info
2015-04-14 12:24:49 +02:00
Marco Ratto
ffc4b7480c
- improved plots with MATLAB annotations [trapped in octave] and different colors
...
- use uimenu to get the name of the observable associated to each CDF from MATLAB plots
[trapped in octave]
- also checks multiple significant parameters by individual observable
2015-04-14 12:16:50 +02:00
Marco Ratto
a44b3fbfe7
Bug fixed for incomplete stock matrices of log-posterior kernel and steady state
...
[used in gsa/filt_mc_.m]
2015-04-14 12:10:46 +02:00
Marco Ratto
e32cbf626f
Fixed bug when neighborhood_width is used : MC option must be forced.
2015-04-14 12:06:25 +02:00
Stéphane Adjemian (Charybdis)
355ed73240
Merge branch 'marco_diffuse_filter_and_smoother_2'
2015-04-14 12:03:07 +02:00
Stéphane Adjemian (Charybdis)
b49388891f
Moved README file for extended preprocessor in subfolder.
2015-04-14 11:57:28 +02:00
Houtan Bastani
44bb0672d2
preprocessor: method out of order
2015-04-14 11:33:02 +02:00
Houtan Bastani
d8b132e39a
preprocessor: remove commented code
2015-04-09 12:36:07 +02:00
Marco Ratto
ca8f0ea006
Harmonize filters/likelihood with smoothers by using new option diffuse_kalman_tol
2015-04-08 15:49:12 +02:00
Marco Ratto
fe48de4406
Re-set default of new option diffuse_kalman_tol to 1e-6:
...
former hardcoded value was 1e-6 in the univariate smoother (i.e. the one most often used).
Updated documentation.
2015-04-08 15:44:39 +02:00
Houtan Bastani
104e4767a4
update license file, #874
2015-04-07 12:01:23 +02:00
Johannes Pfeifer
aea49e2fa0
Add unit test for mode_computations
2015-04-06 12:13:14 +02:00
Johannes Pfeifer
49d4643481
Fix copy and paste typo in gmhmaxlik.m
2015-04-06 12:04:11 +02:00
Johannes Pfeifer
d1a80acb44
Replace non-available simulated annealing by Matlab's routine
...
Closes #814
2015-04-06 11:28:32 +02:00
Johannes Pfeifer
39ed3eb3f0
Fix bugs related to allowing mode_compute to be a string
...
Crashes otherwise
2015-04-06 11:24:42 +02:00
Johannes Pfeifer
a81c9da9a2
Add simulated_annealing.m to optimizers
...
Restores compatibility of mode_compute=102
2015-04-06 11:08:20 +02:00
Johannes Pfeifer
405b1f7368
Add solvopt.m to Dynare optimizers
...
Restores compatibility of mode_compute=101
2015-04-06 11:05:12 +02:00
Johannes Pfeifer
c46b6cdfea
Clarify Ramsey optimal policy
2015-04-04 10:45:17 +02:00
Houtan Bastani
23b322d6c6
Merge pull request #866 from JohannesPfeifer/mode_check
...
Remove hard-coded coloring of legend in mode_check.m
2015-04-03 18:07:14 +02:00
Stéphane Adjemian (Charybdis)
6aa5ceabb2
Fixed bug introduced in 363c4a61222103fda22adff9bdb1115a6073428a.
2015-04-03 18:02:03 +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
Marco Ratto
e97e5c3407
Exclude zero columns of T from Kitagawa transformation, since ordschur is extremely noisy for multiple zero eigenvalues.
...
This can make a lot of difference for large models that have hundreds of definitions.
2015-04-03 18:02:03 +02:00
Marco Ratto
5297836577
Harmonize criteria for exiting diffuse steps in likelihood with the smoother.
...
Since initial Pinf is well scaled to unity, crit1= 1.e-6 is used for smoother and should also apply to likelihood evaluations.
2015-04-03 18:02:03 +02:00
Marco Ratto
7c838b133f
Use local variables for restrict var list, to avoid overwriting the global and messing likelihood evaluations later on.
2015-04-03 18:02:03 +02:00
Johannes Pfeifer
8cc9fbd69a
Remove hard-coded coloring of legend in mode_check.m
...
Now displays correct colors with new Matlab plot interface (>2014a)
2015-04-02 20:52:49 +02:00
Stéphane Adjemian
f9154273d0
Merge pull request #854 from JohannesPfeifer/smoother_var_initialization_fix
...
Fix size of filtered variables matrix
2015-04-01 17:59:12 +02:00
Stéphane Adjemian (Charybdis)
f3d1a6c68e
Merge branch 'master' into marco_analytic_derivation
2015-04-01 16:41:38 +02:00