1. 07 Nov, 2017 1 commit
  2. 09 Oct, 2017 1 commit
  3. 06 Oct, 2017 2 commits
  4. 05 Oct, 2017 1 commit
  5. 30 Aug, 2017 1 commit
  6. 28 Aug, 2017 2 commits
  7. 24 Aug, 2017 2 commits
  8. 07 Jul, 2017 1 commit
  9. 23 May, 2017 1 commit
  10. 27 Apr, 2017 1 commit
  11. 26 Apr, 2017 1 commit
  12. 04 Apr, 2017 1 commit
  13. 03 Apr, 2017 1 commit
    • Stéphane Adjemian's avatar
      Added option nonlinear_filter_initialization. · f9a462bf
      Stéphane Adjemian authored
      Default value is 1 (initialization with the ergodic variance of the reduced
      form solution of the model approximated at order one).
      
      If the model has unit roots, the user must use `nonlinear_filter_initialization=3`,
      which select an identity matrix for the initial covariance matrix of the state variables.
      
      A side effect of this option is to temporarily change the value of options_.qz_criterium to
      a value above one (ie 1+1e-6) so that the unit roots are not rejected. If the
      model has unit roots and if the and if the option
      nonlinear_filter_initialization has a value less than 3, the evaluation of the
      likelihood will fail, because by default the unit root is counted as an
      unstable root.
      f9a462bf
  14. 31 Mar, 2017 2 commits
  15. 28 Mar, 2017 1 commit
  16. 24 Mar, 2017 1 commit
  17. 21 Mar, 2017 3 commits
  18. 08 Feb, 2017 1 commit
    • Stéphane Adjemian's avatar
      Removed penalty_hessian routine. · 9fbef0c1
      Stéphane Adjemian authored
       + 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 2446ab02.
      
      An option is available for computing the hessian with the penalized
      objective function.
      9fbef0c1
  19. 27 Jan, 2017 1 commit
  20. 20 Jan, 2017 1 commit
    • Stéphane Adjemian's avatar
      Fixed bug. · da31cca7
      Stéphane Adjemian authored
      NO_POSTERIOR_KERNEL_DENSITY option was not modifying the expected field in
      options_, consequently this option was no honored.
      da31cca7
  21. 27 Dec, 2016 1 commit
  22. 26 Dec, 2016 2 commits
  23. 19 Dec, 2016 1 commit
  24. 17 Dec, 2016 2 commits
  25. 01 Dec, 2016 1 commit
  26. 04 Nov, 2016 1 commit
  27. 06 Oct, 2016 1 commit
  28. 02 Oct, 2016 1 commit
  29. 30 Sep, 2016 1 commit
  30. 23 Sep, 2016 1 commit
  31. 14 Sep, 2016 1 commit
  32. 21 Jul, 2016 1 commit