503b055df2
Fixed bugs in varlist_indices.m (thanks to Johannes Pfeifer for pointing them)
2011-01-10 13:11:25 +01:00
05b9c70ae8
removing command CALIB: it has not been working for a long time. Calibration is in fact a special case of method of moments and should be replaced by such a method.
2011-01-04 10:16:38 +01:00
256ff761e8
stochastic simulations: fix the number of generated data to be exactly options_.periods
...
added ./tests/simul
2011-01-02 16:55:15 +01:00
35765ee01c
- Matlab wrappers for the "print" option of bytecode
2010-12-31 17:03:19 +01:00
aab5b2a488
- Manages the temporary terms when the model is evaluated block by block
2010-12-31 16:22:24 +01:00
129303430e
The direction is not computed in same way in the dynamic case and in the static case
2010-12-31 16:20:35 +01:00
a882a25760
- length(args) is used instead of size(args, 2) to avoid to transpose column vectors
2010-12-31 16:16:29 +01:00
2dbeddd1a9
- Manages the temporary terms when the model is evaluated block by block
2010-12-31 16:00:49 +01:00
2707965da1
- extends the compatibility of dr1 with bytecode
2010-12-31 15:56:49 +01:00
7209f3638a
- consider a non sparse matrix for the Jacobian for compatibility reasons with dynare_solve
2010-12-31 15:53:28 +01:00
dcc46959b4
The default for options_.qz_criterium is now different for different contexts:
...
qz_criterium = 1+1e-6
- stoch_simul
- osr
- check
- dynare_estimation_1 with lik_init =2 or 3
qz_criterium = 1+1e-6 for dynare_estimation_1 with lik_init = 1
This correct a bug in estimation and computation of the posterior distribution of moments of endogenous variables when the priors permit roots too close to 1.
2010-12-23 08:34:13 +01:00
6bb8d41909
Revert "- added a test an a penalty in estimation (DsgeLikelihood.m) if, in a stationary model (lik_init==1), a particular parameter set generates unit roots."
...
There is a better way of dealing with occasional non-stationary models in estimation
This reverts commit 8c0fb55206
2010-12-22 09:40:39 +01:00
8c0fb55206
- added a test an a penalty in estimation (DsgeLikelihood.m) if, in a stationary model (lik_init==1), a particular parameter set generates unit roots.
...
- modified lyapunov_symm to return absolute value of unit roots in a third argument
2010-12-21 21:26:57 +01:00
ef56f87796
- fix for Octave printf
...
- missing semicolon
2010-12-20 15:55:46 +01:00
55e553f6dc
Missing semi-colons
2010-12-20 15:49:55 +01:00
ac0c0a935e
Cosmethics
2010-12-17 09:22:12 +01:00
b186eb7e84
Fixes for console mode in serial and parallel execution
2010-12-17 09:21:30 +01:00
191de78310
Introduced console mode monitoring of parallel runs;
...
Minor changes
2010-12-17 09:20:30 +01:00
916caf2e4b
Almost entirely rewritten:
...
1) got rid of recursive form;
2) better help
3) simplified structure and split between errors and warnings.
2010-12-17 09:16:30 +01:00
c2f0037e55
small addendum to help
2010-12-17 09:14:35 +01:00
d17faaf876
cosmethics
2010-12-17 09:12:54 +01:00
41c59778cd
For slave processes on the local machine, impose the same dynare path as the master.
2010-12-17 09:11:11 +01:00
c0cb3dd4fb
Compatibility fix for Octave: under Octave, union() called on a column vector and a scalar returns a row vector, hence the need to add the 'rows' option
2010-12-16 16:14:23 +01:00
b63590712c
fixed bug in argument number check and removed unused variable
2010-12-14 18:42:29 +01:00
e04a07c771
changed variable name to make clear that, for multinormal variables, we are using the Cholesky decomposition of the covariance matrix.
2010-12-11 18:09:01 +01:00
c1cb452e3d
correcting headers of all Kalman filter functions: returns MINUS loglikelihood
2010-12-10 22:31:30 +01:00
7a09356423
DsgeLikelihood.m header correction: the function returns MINUS log posterior
2010-12-10 22:28:06 +01:00
dfbd10a524
added choice of parameter set to shock_decomposition. This corrects a bug when shock_decomposition is requested after Metropolis iterations.
2010-12-08 13:02:59 +01:00
ea86e8cefc
adding function evaluate_smoother that calls DsgeSmoother with a given type of parameter set
2010-12-08 13:02:59 +01:00
9b166fdb08
NaN parameters warning: remove \n, it is not interpreted in a warning context (at least under Octave)
2010-12-08 11:32:45 +01:00
03fac30723
remove get_innovation_contemporaneous_impact.m that isn't called by the preprocessor or by any other function
2010-12-07 13:43:17 +01:00
3af62c4ac9
Fixed bug in resid.m (order matters in "or" statement, because of short-circuit: "ind" is not defined if "istag" is false)
2010-12-06 18:24:08 +01:00
0d1fd700c7
Posterior distribution of conditional variance decomposition: storing the steps chosen by the user in a separate field.
2010-12-05 10:47:31 +01:00
53d391785d
corrected bug in get_name_of_the_last_mh_file.m
2010-12-01 09:57:08 +01:00
d4a908dda1
Removed useless input argument.
2010-11-30 10:02:37 +01:00
116bf69fbf
Seed: fix bug under Octave and MATLAB < 7.7
2010-11-29 11:23:59 +01:00
1fc8bbd6d3
bugs correction in computation of posterior moments for
...
-conditional variance decomposition
-hpdsup
-moments with no variance in their posterior distribution
modification of computation of conditional variance decomposition
2010-11-28 11:03:40 +01:00
73fb71b9c8
Removed references to options_.simul_seed.
2010-11-27 00:19:22 +01:00
04469ddce1
- Add variable input arguments
...
- Correct the error condition in A_times_B_kronecker_C
2010-11-26 17:46:02 +01:00
db54c2d76e
Added a routine for setting the random number generator and the seed, independantly of the version of matlab. Removed the calls to rand('state',x) in simult.m and added a test file. Still need to adapt the routines related to mcmc and swz.
2010-11-26 17:16:12 +01:00
618f72bc91
resid displays the name of the equations if they are specified in equations tags.
2010-11-25 17:31:11 +01:00
ffaec7ba89
- First elements needed for stochastic simulation of a block decomposed model
2010-11-25 16:11:27 +01:00
61a251535f
Correction of bugs in check command with block decomposition and
...
Add the missing dr_block m-file
2010-11-25 16:10:46 +01:00
b356c5e58e
add_auxiliary_variables_to_steadystate is compatible with bytecode option
...
+ minor corrections in steady_.m
2010-11-25 16:00:54 +01:00
1626b0aa76
Added specialized code for purely forward models.
2010-11-25 14:32:33 +01:00
1eccfb311a
Remove obsolete files
2010-11-25 13:10:10 +01:00
6c22d398e6
Perfect foresight simulation: clean up handling of warnings
2010-11-24 19:01:46 +01:00
9d4385da5d
Put back M_.maximum_lag.
2010-11-19 16:30:55 +01:00
ccf778e63a
Efficiency improvements in first order stochastic simulations (Simulate data in deviation to the steady state and then,
...
after the loop, add the steady state using bsxfun routine, which is parallelized in recent version of matlab). Removed
use of M_.maximum_lag.
2010-11-18 14:55:07 +01:00
e614227c3f
Added warning message when some of the deep parameters are not initialized.
2010-11-17 17:09:39 +01:00
d60c32e6a2
Fixed bug in dealing with _steadystate functions that modify deep parameters. See trac ticket #103 .
2010-11-15 14:35:21 +01:00
5c709e4733
Changed set_dynare_threads.m. The number of threads in parallelized mex files to be used if dynare is built with the openmp flag
...
(--with-openmp) is not passed by an environment variable anymore. The function set_dynare_threads changes the default value of the
number of threads (default is 1) in the options_.threads structure. Changed calls to sparse_hessian_times_B_kronecker_C and
A_times_B_kronecker_C dlls accordingly.
2010-11-15 13:06:39 +01:00
37f14e9bc9
Removed unused routines for (diffuse) kalman filter evaluations.
2010-11-12 17:20:02 +01:00
31b734e9b3
Do not print residuals of the static equations for the equations of the auxiliary variables.
2010-11-04 14:19:28 +01:00
88dc361ba1
unix info added
2010-10-29 12:30:41 +02:00
5763721a71
almost complete version of testing routine
2010-10-29 12:28:44 +02:00
23ba760c3b
Implement solve_algo=0 for Octave ( closes #144 )
2010-10-28 12:07:50 +02:00
67c28ea5f3
PosteriorIRF: added missing ".mat" extension (necessary for Octave)
2010-10-28 10:39:53 +02:00
0730213059
Fix end of line convention and copyright years
2010-10-27 18:35:42 +02:00
1aceb504fd
Check command is now compatible with block and bytecode options
2010-10-27 15:34:48 +02:00
9cebf270e5
- correction of informations displayed in Solve_one_boundary
2010-10-27 15:30:00 +02:00
58a8988ef3
Moved another parallel routine previously left out.
2010-10-27 11:27:10 +02:00
b5ebff4ff4
bug fix for parallel execution
2010-10-27 11:24:23 +02:00
7ef63899b1
add paralle_info input to Analize;
...
fix spelling of function names
2010-10-27 11:05:39 +02:00
5915f8a059
Fixed misspelled routines!
2010-10-27 11:03:21 +02:00
0a38f8ac3b
Initialize here default of MatlabOctavePath;
...
implies using global options_ in place of local argument
2010-10-25 09:53:09 +02:00
fe8724d26c
Merge remote branch 'ratto/master'
2010-10-22 17:22:37 +02:00
5059c48460
solve_algo=7 (GMRES) is forbidden under Octave
2010-10-22 16:52:02 +02:00
39718c8645
- extends steady algorithms to solve_algo=5, 6, 7 and 8 for a block decomposed model without bytcode
...
- solve_algo is reordered =
* 0: fsolve
* 1: solve1
* 2, 4: solve1 + block decomposition
* 3: csolve
* 5: bytecode own solver (use Gaussian elimination + sparse matrix)
* 6: LU decomposition with UMFPack (method handling sparse matrix in Matlab)
* 7: GMRES
* 8: BiCGStab
2010-10-22 16:49:47 +02:00
f336a68ecc
Update tests for values of solve_algo and stack_solve_algo
2010-10-22 13:00:49 +02:00
a46969f9fb
Deleted buggy fclose instruction
2010-10-22 11:40:29 +02:00
d8121365dd
1) Renamed options according ti WiKi table;
...
2) get rid of MasterName and DyMo;
3) more comments;
2010-10-22 11:27:26 +02:00
f915cbe92f
- Minor corrections in wrapper to compute the steady-state
2010-10-22 11:08:47 +02:00
1a09426706
- Correction of several bugs with Octave
...
- Add wrapper needed to compute the steady-state using bytecode and solve_algo = 0, ..., 4
2010-10-22 10:09:07 +02:00
2239d9f1cd
Fixed bug in diffuse filter
2010-10-21 15:43:30 +02:00
81823ad035
New oct-file for "ordschur": the diffuse filter now works under Octave
2010-10-21 15:43:13 +02:00
e8a193bbfb
Track #21 : Add partial information: Correcting bug stemming from the differences in QZ calculation under Octave and Matlab: modified PI_gensys to use C.Sims' Lapack zgges based qzcomplex() qz decomposition routine under Octave which gives nearly identical complex results to Matlab QZ (except for transpose of q matrix and few sign changes in the output matrices)
2010-10-20 11:50:06 +01:00
567c5bcb1f
Fixed bug in the initialization of the metropolis when mode_compute=6 was previously used (so that we have an optimal mh scale factor)
...
and option load_mh_file is used. After the (stochastic) optimization, the optimal value of the scale parameter is saved in a mat file.
2010-10-18 14:39:48 +02:00
de490e9e4a
delete remnants of previous Bayesian irf computations
2010-10-15 11:36:26 +02:00
c15508eae0
1) better display of remote errors;
...
2) fixed bug in sync of sequence of files sent to remote slaves;
2010-10-15 11:33:40 +02:00
6612a7f0bc
Trap introduced for windows was a bug for unix
2010-10-15 11:31:26 +02:00
6d2f9dc6d8
Added missing case in print_info
2010-10-14 12:03:36 +02:00
609293b398
dr1.m: move up error message for purely forward models, otherwise it is not displayed with k_order_perturbation
2010-10-14 11:36:13 +02:00
153f627dfe
Explicit error message for 2nd and 3rd order purely forward models
2010-10-13 18:52:13 +02:00
505ed643a1
Fix for computing 1st order irf of purely forward models
2010-10-13 18:52:07 +02:00
7a983a3105
sim1.m, bksup1.m: removed global variables, other minor cleanups
2010-10-13 18:51:56 +02:00
f30d6858e6
simk.m: minor cleanups
2010-10-13 18:40:21 +02:00
b4d11d399c
Resurrect simk.m for purely forward models
2010-10-13 18:38:09 +02:00
5f0dac24bf
Merge remote branch 'ferhat/master'
2010-10-12 11:40:39 +02:00
cc70913b7d
Properly close waitbar in serial computations.
2010-10-12 08:07:57 +02:00
f079411126
bug fix: precisely determine from which machine to get the output files
2010-10-11 20:17:03 +02:00
6a1826aa53
1) fixed bug introduced moving folders (parallel routines no longer on path!);
...
2) fixed bug with logic of islocal;
3) better cleanup of remnants;
4) fixed bug with sending and DELETING P*End.txt files;
2010-10-11 19:40:42 +02:00
badda23160
fixed bug when one slave is local and others are remote
2010-10-11 19:37:43 +02:00
8f36437662
- Extends the algorithms compatible with bytecode to compute the steady-state (ticket #11 ). The new values for solve_algo are:
...
* 0: fsolve
* 1: solve1
* 2, 4: solve1 + block decomposition
* 3: csolve
* 5: LU decomposition with UMFPack (method handling sparse matrix in Matlab)
* 6: GMRES
* 7: BiCGStab
* 8: bytecode own solver (use Gaussian elimination + sparse matrix)
- Bytecode can now evaluate a specific block instead of the overall blocks (new bytecode's option 'Block = block_number')
2010-10-11 19:21:32 +02:00
75dd6f0d7b
Merge remote branch 'ratto/master'
2010-10-11 17:24:07 +02:00
a5b8819518
Partial information: added missing copyright notice
2010-10-11 17:16:52 +02:00
9abc57a195
added missing utility for identification
2010-10-11 17:14:49 +02:00
44b268ec67
Moved qzdiv.m and qzswitch.m to top-level (they are now used by both partial information code and mjdgges.m)
2010-10-11 17:14:33 +02:00
cf9f901ef8
Partial information: moved disclyap_fast to the partial_information subdir
2010-10-11 17:11:05 +02:00
8849dd2bc9
Partial info/Octave compatibility fix: ismember() does not work with char arrays under Octave
2010-10-11 16:48:16 +02:00
Sébastien Villemot
Michel Juillard
Ferhat Mihoubi
Marco Ratto
Houtan Bastani
Stéphane Adjemian (Charybdis)
George Perendia