Michel Juillard
76f008e4e4
PATH: remove calls to lcppath that doesn't work with perfect foresight models
2016-05-31 12:10:12 +02:00
Michel Juillard
ddb9c0657c
fixing linear complementarity problem
2016-05-31 12:10:12 +02:00
Michel Juillard
450d7b099a
fixing error status for PATH hook
2016-05-31 12:10:12 +02:00
Michel Juillard
c159dff19c
adding hook for PATH solver
2016-05-31 12:10:12 +02:00
Johannes Pfeifer
465f55bb0e
Correct evaluation of residuals in dynare_solve.m when exitflag does not clearly indicate solution
...
Otherwise, it crashes with bytecode
2016-05-05 10:46:39 +02:00
Johannes Pfeifer
1457666221
Only accept steady state when fsolve retuns with normal return code 1
...
Otherwise, termination due to tolerance criteria will be accepted as correct steady state
2016-03-22 21:55:51 +01:00
Michel Juillard
c3c72192ac
small bug corrections
2015-07-27 16:52:36 +02:00
Michel Juillard
2207b51aaa
correcting typo
2015-07-27 15:51:48 +02:00
Michel Juillard
9b8077fc28
Merge branch 'master' into new_ep
2015-07-21 09:30:58 +02:00
Stéphane Adjemian (Charybdis)
27922a349c
Fixed typo.
...
The value of jacobian_flag was not passed to fsolve.
2015-07-07 17:55:41 +02:00
Stéphane Adjemian (Hermes)
c97aaad62d
Fixed bug.
...
Rename options as options4fsolve, otherwise original options structure is lost.
2015-05-29 15:19:20 +02:00
Michel Juillard
d81dda5044
fixing bug in last commit
2015-05-25 18:52:46 +02:00
Michel Juillard
3ef5bc2989
Merge branch 'master' into new_ep
2015-05-25 17:20:23 +02:00
Michel Juillard
f03839cf12
removed "global options_" in dynare_solve.m and made it an
...
argument. Modified all calling functions.
2015-05-25 17:15:09 +02:00
Michel Juillard
a53385ea28
dynare_solve: fix initial check for models with complementarity conditions
2015-03-07 10:10:35 +01:00
Stéphane Adjemian (Charybdis)
0851d277c4
Use correct tolerance parameter when solving perfect foresight model
...
with stack_solve_algo=7.
2015-02-18 23:52:17 +01:00
Michel Juillard
9fa92b0d02
fixing error message
2014-11-14 07:15:24 +01:00
Stéphane Adjemian (Charybdis)
ce97a27c9d
Cosmetic changes.
2014-09-10 17:49:33 +02:00
Johannes Pfeifer
308ce25031
Make sure that solve_algo=0 inherits options from options_.steady
...
Otherwise, MaxIter is hard-coded
(cherry picked from commit 2499f641ac2fab7ddf938263d0ee34f73765c4c8)
2014-09-10 17:49:33 +02:00
Sébastien Villemot
1a4df2fedf
Merge pull request #659 from JohannesPfeifer/steady_fix
...
Fix size of NaN vector returned so that resid.m does not crash
2014-06-04 16:59:14 +02:00
Michel Juillard
979c1af53b
perfect foresight with lmmcp: removed display of iterations
2014-05-28 09:23:10 +02:00
Johannes Pfeifer
616dad634a
Fix size of NaN vector returned so that resid.m does not crash
2014-05-26 16:16:22 +02:00
Michel Juillard
5425245ec1
adding lmmcp
2014-05-18 21:49:52 +02:00
Sébastien Villemot
e8ac5da9ea
Fix trust region on entire model.
2014-05-09 15:33:26 +02:00
Michel Juillard
57ae7d7fd9
adding solve_algo=9: using trust_region algorithm on the entire model
2014-05-07 21:25:10 +02:00
Johannes Pfeifer
d9d4fef251
Move check of Jacobian matrices for NaN and Inf to dedicated function for reuse in model_diagnostics.m
2014-03-28 14:52:06 +01:00
Sébastien Villemot
a5be9dd845
Add trust-region nonlinear solver.
...
Closes #260
2014-02-04 17:56:27 +01:00
Sébastien Villemot
3fe5a728c6
Remove bad_cond_flag argument from solve1.
...
This flag was not grounded on any solid theoretical foundation.
This commit actually makes solve_algo=2 to be exactly the same than solve_algo=4.
2014-02-04 17:56:27 +01:00
Stéphane Adjemian (Charybdis)
ddd95361ba
Cosmetic changes. Added isoctave function to replace exist('OCTAVE_VERSION').
2013-11-04 10:54:45 +01:00
Stéphane Adjemian (Charybdis)
969fe3c209
Renamed options_.solve_maxit as options_.steady.maxit.
2013-10-09 15:38:07 +02:00
Johannes Pfeifer
aa59a4e173
Cosmetic fix to displaying steady state error
...
Gets rid of white space
2013-07-18 15:00:41 +02:00
Johannes Pfeifer
ab1cb25ed3
Filter out cases where invalid starting value leads to NaN or Inf in Jacobian.
...
Replaces cryptic message "Some element of Newton direction isn't finite. Jacobian maybe singular or there is a problem with initial values" with more expressive message.
2013-06-21 10:37:16 +02:00
Michel Juillard
2e59409df4
corrected bug introduced in recent commit 48e00c5789
2012-11-01 09:26:05 +01:00
Michel Juillard
48e00c5789
corrected and clarified error messages when steady state computation fails
2012-10-31 11:09:25 +01:00
Michel Juillard
70e162c736
removed globals from solve1.m
2012-08-06 23:32:40 +02:00
Sébastien Villemot
edbf3bd175
Remove useless calls to set_default_option
2012-04-20 17:08:11 +02:00
Stéphane Adjemian (Charybdis)
323e6fdf4f
Added a second gstep option used as a parameter for the routine computing the hessian matrix.
...
In some cases, for instance for the non linear filters, it helps to reduce this new gstep parameter
to get a positive definite hessian matrix. options_.gstep is now a 2*1 vector. The first element is
the old gstep parameter, the second element is the new gstep parameter. The step defined for the
computation of the hessian matrix is now:
h1=max(abs(x),sqrt(gstep(1))*ones(n,1))*eps^(1/6)*gstep(2);
2012-03-09 16:36:26 +01:00
Houtan Bastani
dcaed91f66
fix to license call (checkout only works for Matlab version >= 2011a)
2012-01-09 11:57:47 +01:00
Houtan Bastani
9be84bdf65
make check: fix bug in check for license
2011-12-23 18:22:41 +01:00
Marco Ratto
dde9d2bfee
license('test',' ... ') gives 1 even if the license exists but it is not active, implying a crash of the dynare matlab session.
2011-10-26 16:27:09 +02:00
Sébastien Villemot
0155b6832e
Remove extra argument (in solve_algo=0 under Octave)
2011-09-19 16:41:53 +02:00
Sébastien Villemot
c07a915d6e
Improve the way we test for existence of toolboxes (more reliable in the case where the toolbox is installed but not activated)
2011-09-19 16:41:53 +02:00
Michel Juillard
6662aa9e26
expanded error message when initial values are incompatible with steady state equations
2011-07-14 12:20:36 +02:00
Sébastien Villemot
02652f6eb8
Updated copyright notices
2011-02-04 17:27:33 +01:00
Sébastien Villemot
757a9067c0
Global reindentation of MATLAB files
2011-02-04 17:17:48 +01:00
Ferhat Mihoubi
ce07223628
- The Temporary terms management with bytecode when the model is block decomposed and solve_algo<5 is now compatible with octave
...
- Gets rid of warning message during the compilation of bytecode
2011-01-14 19:24:18 +01:00
Ferhat Mihoubi
708a062338
The Octave version of fsolve do not converge when it starts from the solution. Check if the initial guess is equal to the solution.
2011-01-14 19:24:17 +01:00
Sébastien Villemot
23ba760c3b
Implement solve_algo=0 for Octave ( closes #144 )
2010-10-28 12:07:50 +02:00
sebastien
502e3e1df8
Beautified MATLAB code (Unix newline convention + Emacs indentation), except: AIM, swz, particle
...
git-svn-id: https://www.dynare.org/svn/dynare/trunk@3250 ac1d8469-bf42-47a9-8791-bf33cf982152
2009-12-16 17:17:34 +00:00
sebastien
fec9486fb9
Change to option solve_algo=4 of steady:
...
Even when the Jacobian is very badly conditioned, continue to use a Newton step
git-svn-id: https://www.dynare.org/svn/dynare/trunk@2862 ac1d8469-bf42-47a9-8791-bf33cf982152
2009-08-21 12:05:34 +00:00