Commit Graph

10174 Commits (feee6f8b7a518a51431c9377b35f358bac5ab1e7)

Author SHA1 Message Date
Stéphane Adjemian (Charybdis) feee6f8b7a Fixed typo. 2017-03-15 17:24:45 +01:00
Stéphane Adjemian (Charybdis) 284df20445 Closes #1403. 2017-03-15 16:51:59 +01:00
Stéphane Adjemian d0ab90275b Merge pull request #1402 from JohannesPfeifer/prior_asymptote
Add warning if prior has an asymptote
2017-03-15 16:33:10 +01:00
Stéphane Adjemian (Charybdis) 1ff816d7bf Updates particles submodule. 2017-03-15 16:30:29 +01:00
Stéphane Adjemian (Charybdis) 4f3b35acba Account for Matlab 2017a.
(cherry picked from commit f3a1619040fbb831830791424015d3d3bf8bdbae)
2017-03-13 23:42:48 +01:00
Houtan Bastani ea13b12cf2 preprocessor: add lag info to equation cross references. closes #1398 2017-03-10 16:59:52 +01:00
Johannes Pfeifer 87fc09863d Add warning if prior has an asymptote
Can happen with beta and gamma distribution. Closes #1401
2017-03-10 15:16:33 +01:00
Houtan Bastani 38f36091a9 test suite: increase timeout option for downloading files to 30 seconds 2017-03-08 12:30:58 +01:00
Houtan Bastani 7341ca4e6d Merge pull request #1399 from JohannesPfeifer/dname
Fix check for existence of dname variable
2017-03-07 13:58:29 +01:00
Johannes Pfeifer 56edaf5cb0 Fix check for existence of dname variable 2017-03-07 13:55:29 +01:00
Houtan Bastani 33b38ab643 preprocessor: write error messages to cerr instead of cout, replace \n with endl 2017-03-01 12:44:31 +01:00
Houtan Bastani 2d25858886 preprocessor: fix bug that caused temporary terms not to be printed for the parameter derivatives of the static model. closes #1397 2017-03-01 12:20:55 +01:00
Houtan Bastani 0b52b778ed preprocessor: for consistency, use equations.size() instead of equation_number() in classes that have equations as a field 2017-02-24 11:20:54 +01:00
Stéphane Adjemian (Charybdis) 927b813ed8 Removed printing of debugging info.
Introduced in ffebf6b262.
2017-02-15 13:35:33 +01:00
Stéphane Adjemian (Charybdis) ffebf6b262 Check positive steady state for loglinear option...
1e-9 is too far from zero. Set 1e-16 as a threshold value for the positivity test.
2017-02-15 12:41:56 +01:00
Stéphane Adjemian (Charybdis) 8925e7ca1f Cosmetic patch (remove some warnings about potentially uninitialized variables in bytecode). 2017-02-13 19:22:26 +01:00
Houtan Bastani 39855c95d2 preprocessor: aesthetic fix 2017-02-10 12:47:43 +01:00
Houtan Bastani fabca90345 preprocessor: aesthetic fix 2017-02-09 16:23:42 +01:00
Houtan Bastani 7ced4b5580 preprocessor: change error to warning because we don’t end processing in this situation; fix error message 2017-02-09 16:21:18 +01:00
Houtan Bastani edd2e98086 preprocessor: replace exit(1) with exit(EXIT_FAILURE) 2017-02-09 12:46:37 +01:00
Stéphane Adjemian (Charybdis) 593d62a7c1 Removed optimizers/fs2000_12.mod from testsuite...
Because the Global Optimization Toolbox is not installed on dynbot@sedna.
2017-02-08 14:29:34 +01:00
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
Stéphane Adjemian (Charybdis) b14cf33c8f Changed priors in tests/optimizers/fs2000_* integration tests.
To avoid asymptote at zero on the autoregressive parameter rho.
2017-02-08 13:07:25 +00:00
Stéphane Adjemian (Charybdis) 5b73d7d59c Added new optimization routine (particleswarm).
Only available under Matlab if the Global Optimization Toolbox is installed.
2017-02-08 13:07:25 +00:00
Stéphane Adjemian (Charybdis) 35078e13ee Completed header + cosmetic changes. 2017-02-08 13:07:25 +00:00
Stéphane Adjemian (Charybdis) 8d4abe4a9f Cosmetic changes. 2017-02-08 13:07:24 +00:00
Stéphane Adjemian (Charybdis) e71e89bb38 Added check on Octave version. 2017-02-08 13:07:24 +00:00
Houtan Bastani e88739a5ea doc: add preprocessor copyright info 2017-02-01 18:16:51 +01:00
Houtan Bastani 25e9a81b71 doc: update for preprocessor 2017-02-01 18:06:44 +01:00
Houtan Bastani 3776b910a1 preprocessor: remove introduction to C++ programming 2017-01-31 17:11:28 +01:00
Houtan Bastani 5b6a888d7f preprocessor: fix typo 2017-01-31 17:11:03 +01:00
Houtan Bastani f1d607af99 preprocessor: remove extra exogenous variables. closes #841 2017-01-27 22:48:17 +01:00
Houtan Bastani 3a0b9a1143 preprocessor: don’t automatically declare external functions 2017-01-27 20:36:26 +01:00
Houtan Bastani 3d4619fd85 preprocessor: handle undeclared external functions and lagged variables 2017-01-27 20:36:26 +01:00
Houtan Bastani cfae273038 keep track of all times an undeclared variable is encountered 2017-01-27 20:36:26 +01:00
Houtan Bastani bfd239aa26 preprocessor: issue warnings for undeclared model variables when the end of the model block is encountered. #1286 2017-01-27 20:36:26 +01:00
Stéphane Adjemian (Lupi) 1cabe76b06 Do not compute posterior kernel densities in some integration tests.
These data are not used in the tests, and take a lot of space in RAM and on
the disk (when the results, e.g. oo_, are saved in <MOD_FILE_NAME>_results.mat).
2017-01-27 15:15:55 +01:00
Stéphane Adjemian 5285f581e8 Merge pull request #1383 from JohannesPfeifer/readme
readme.md: Fix typos and add info on make clean
2017-01-27 15:09:48 +01:00
Johannes Pfeifer f9f755770d readme.md: Fix typos and add info on make clean and build dependencies 2017-01-27 14:28:19 +01:00
Stéphane Adjemian (Charybdis) 62c954ff12 Fixes #1287. 2017-01-25 22:13:36 +01:00
Stéphane Adjemian (Charybdis) 40993c4ebe Document integration and unit tests in README.md. 2017-01-25 17:14:51 +01:00
Stéphane Adjemian (Charybdis) 0b1b873160 Updated dseries submodule.
Fixes remaining Octave failing unit tests.
2017-01-25 09:18:28 +01:00
Stéphane Adjemian (Charybdis) b152f3920d Added the possiblity to run the testsuite on a subset of tests.
*Example*

To run all the tests related to the numerical gradient, just type:

~$ make m/gradient

in the tests subfolder. These tests will be done with Matlab. If you want to
test these routines against Octave instead:

~$ make o/gradient

If you want to test against Matlab and Octave:

~$ make gradient.

*Remark*

[1] Integration tests in subfolders can be accessed using the same approach. For
instance:

~/m/observation_trends_and_prefiltering/MCMC

will run all the mod files with matlab in tests/observation_trends_and_prefiltering/MCMC.

[2] Autocompletion works.

[3] Some folders/subfolders are still missing...

(cherry picked from commit 0271f2e8c13d5fe94871fa7aed027047a2cfd36f)
2017-01-24 16:00:50 +00:00
Houtan Bastani a6e74ac7af doc: update manual copyright date 2017-01-23 14:24:30 -05:00
Stéphane Adjemian (Charybdis) da31cca7d4 Fixed bug.
NO_POSTERIOR_KERNEL_DENSITY option was not modifying the expected field in
options_, consequently this option was no honored.
2017-01-20 17:35:43 +01:00
Houtan Bastani 1de5ac8ecf Merge pull request #1379 from JohannesPfeifer/octave_tests
Adjust Octave unit tests
2017-01-18 09:06:40 -05:00
Johannes Pfeifer 875497bc71 Unit test lpdfgweibull.m: adjust upper integration bound under Octave
Does not correctly integrate otherwise as more than 99% of support is 0
2017-01-18 11:01:13 +01:00
Johannes Pfeifer ce4ab7f7d8 Unit test wblinv.m under Octave: adjust tolerance 2017-01-18 10:59:53 +01:00
Houtan Bastani 477a09e9ef Merge pull request #1376 from JohannesPfeifer/remove_unused_varexo2
unit tests: remove more unused exogenous variables
2017-01-11 00:26:26 +01:00
Johannes Pfeifer d5a6f132b5 unit tests: remove more unused exogenous variables 2017-01-10 19:58:44 +01:00