Marco Ratto
93e7be66e7
proper use on diffuse_kalman_tol in univariate diffuse filter and smoother
2015-10-13 17:28:29 +02:00
Marco Ratto
05fc096569
Make multivariate kalman filter and smoother robust to badly scaled covariance matrix of observables.
...
This avoids shifting to univariate filter in most cases.
2015-10-13 17:15:01 +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
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
Stéphane Adjemian (Scylla)
89fa0ae9b6
Fixed typo.
2014-09-11 17:33:42 +02:00
Marco Ratto
a6bddb2d57
Improve computation of outer product gradient for univariate Kalman algorithms, by exploiting the larger number of individual densities computed during recursions (used in optimizer number 5).
2014-07-23 16:33:39 +02:00
Sébastien Villemot
61485ab809
Fix copyright notices
2013-06-12 17:04:46 +02:00
Johannes Pfeifer
1df8bf15c2
Bugfix in rplot + typo correction
2013-03-18 10:59:32 +01:00
Michel Juillard
bbd95b3a60
adding Ed Herbst fast implementation of the Kalman filter and test
...
cases with timing. Still needs preprocessor interface (option) and documentation.
2013-02-21 17:47:16 +01:00
Marco Ratto
e637319be5
bug fix. the diffuse filter should simply penalize the likelihood with the NaN without breaking the estimation (this is in line with missing_observations_kalman_filter_d.m).
2012-09-14 17:05:35 +02:00
Marco Ratto
6b91301365
Reduce memory requirements for analytic Hessian.
...
Tested with QUEST III (Ratto et al. 2009): 63 params and 59 state dimension.
2012-08-21 16:00:55 +02:00
Marco Ratto
7683175e8e
Bug fix in terms for outer product gradient with analytic derivatives
2012-08-21 15:45:25 +02:00
Marco Ratto
f7aae77f16
removed useless elements + reduce loops for efficiency
...
(cherry picked from commit f2271264994af253ccdc7f01478320e4d2c2c0e7)
2012-08-13 13:04:34 +02:00
Ferhat Mihoubi
307d5d5d6a
Initializes the s variable
2012-07-01 15:19:36 +02:00
Michel Juillard
2fa433f18e
fixing bug for estimation of models with measurement errors and
...
missing observations
2012-06-19 16:35:44 +02:00
Sébastien Villemot
129553579a
Merge remote-tracking branch 'ratto/master'
2012-06-08 18:24:18 +02:00
Sébastien Villemot
1f9cea669a
Update copyright notices
2012-06-08 18:22:34 +02:00
Marco Ratto
bf88b6e93d
LIKK needs to be initialized when analytic derivation =0.
2012-06-08 17:06:59 +02:00
Marco Ratto
2fecf9946b
1) Extended optimizer = 5 for analytic derivatives;
...
2) Start adapting identification routines to allow computation of analytic asymptotic Hessian with KF routines
2012-06-08 14:23:18 +02:00
Michel Juillard
c15123878f
fixed rare bug in Kalman filter when one should switch to steady state
...
filter in last period of the sample
2012-05-10 11:09:06 +02:00
Marco Ratto
da9ec0f187
Estimation with analytic scores and hessian;
...
This includes re-setting the list of output arguments in objective functions
Added test function
2012-04-29 21:18:33 +02:00
Marco Ratto
459744649d
bug fixes: output argument name "lik" and check of no more missing obs
2012-04-29 21:18:33 +02:00
Michel Juillard
d86daa0169
fixing bug in recent commit 919c2f8fb4
2012-01-23 16:24:47 +01:00
Michel Juillard
919c2f8fb4
correcting bug with presample and diffuse filter + simplified logic
...
for computation of likelihood with presample
2012-01-22 22:40:46 +01:00
Michel Juillard
f0d1f033b0
correcting bug in univariate diffuse filter with presample
2012-01-22 18:36:31 +01:00
Stéphane Adjemian (Charybdis)
ecac871435
Changed the name of DsgeLikelihood (-> dsge_likelihood).
2011-12-26 17:46:48 +01:00
Michel Juillard
976c8c1654
making univariate Kalman filter code simpler and more efficient
2011-12-15 17:35:26 +01:00
Michel Juillard
e4c803d0db
fixed issues with estimation of non-stationary models. Option lik_init=2
...
is contradictory with diffuse_filter or unit_root_variables
declaration. Models with non-stationary variables, but only stationary
observed variables need diffuse_filter option and make a useless call
to kalman_filter_d (this seems better than trying to distinguish these
rare cases)
2011-11-21 12:39:02 +01:00
Michel Juillard
9d91625c10
fixing bug related to measurement errors
2011-11-02 11:10:58 +01:00
Stéphane Adjemian (Charybdis)
88814c9b93
Added texinfo header.
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
c2eb6f11d3
Added texinfo header. Fixed efficiency issue (missing parenthesis).
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
e60032a6f7
More typo fixes.
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
52e803f3b7
Fixed typo.
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
681a7e1df5
Added texinfo header. Fixed efficiency issue (missing parenthesis).
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
fc21af10ea
Cosmetic changes.
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
ffbb5bc765
Added texinfo header.
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
e270c29bd9
Fixed texinfo header (Z was missing in the description).
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Charybdis)
95c3b6396c
Added texinfo header and cosmetic changes.
2011-10-25 12:34:05 +02:00
Stéphane Adjemian (Scylla)
f2ca6d0ad9
Changed kalman filter routines to allow for arbitrary initial conditions (needed for the introduction of breaks on the estimated
...
parameters and also for the estimation of the initial states).
Added specialized routines for steady state kalman filter.
Completed header of DsgeLikelihood (missing refs to the routines called by DsgeLikelihood).
2011-09-19 17:01:24 +02:00
Houtan Bastani
c59133562c
more short-circuit logicals in the place of element-wise logicals
2011-02-22 15:06:38 +01:00
Marco Ratto
13ea1c0046
A better way to compute outer product gradient (for optimizer 5) with the univariate Kalman Filter.
2011-02-14 11:48:29 +01:00
Houtan Bastani
43479f6ef3
use short-circuit ops (|| and &&) as opposed to (| and &) to avoid warnings in Octave (and save time)
2011-02-10 15:54:23 +01:00
Sébastien Villemot
02652f6eb8
Updated copyright notices
2011-02-04 17:27:33 +01:00
Michel Juillard
c1cb452e3d
correcting headers of all Kalman filter functions: returns MINUS loglikelihood
2010-12-10 22:31:30 +01:00
Stéphane Adjemian (Scylla)
bb47d78611
Fixed bug in Kalman filter routines + Cosmetic changes.
2010-06-24 18:20:07 +02:00
Michel Juillard
61e78763da
correcting bug in kalman filter when it reaches the steady state (multivariate diffuse Kalman filter, both versions, regular Kalman filter with missing observations)
2010-06-09 18:00:17 +02:00
Stéphane Adjemian (Scylla)
a512b9b263
Bug fix. We don't have to check the convergence of the Kalman filter to its steady state if the
...
covariance matrix F is singular.
2010-06-09 11:04:59 +02:00
Michel Juillard
4bde063669
Kalman filters: fixing initialization of oldK to Inf instead of 0
2010-04-16 12:20:57 +02:00
Marco Ratto
24d41f66c8
fixed bug with tolerance for steady state kalman filter
2010-03-05 11:26:29 +01:00
Stéphane Adjemian (Charybdis)
3f1fad9cd2
Added trap for cases where diffuse filters fail (replaced error messages by warning messages).
2010-02-05 23:18:08 +01:00