1. 22 Jan, 2021 1 commit
  2. 10 Jan, 2021 1 commit
  3. 16 Dec, 2020 1 commit
  4. 26 Jun, 2020 1 commit
  5. 23 Jun, 2020 1 commit
  6. 17 Jun, 2020 1 commit
  7. 10 Mar, 2020 1 commit
  8. 24 Jan, 2020 1 commit
  9. 20 Dec, 2019 2 commits
  10. 21 Nov, 2019 3 commits
  11. 19 Mar, 2019 1 commit
  12. 13 Nov, 2018 1 commit
  13. 16 May, 2018 1 commit
  14. 01 Feb, 2018 1 commit
  15. 09 Jan, 2018 1 commit
  16. 10 Oct, 2017 1 commit
  17. 09 Oct, 2017 1 commit
  18. 05 Oct, 2017 1 commit
  19. 16 Aug, 2017 2 commits
  20. 01 Jun, 2017 1 commit
  21. 16 May, 2017 2 commits
  22. 05 Apr, 2017 1 commit
  23. 23 Mar, 2017 2 commits
  24. 22 Mar, 2017 1 commit
  25. 17 Mar, 2017 1 commit
  26. 16 Mar, 2017 1 commit
  27. 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
  28. 27 Dec, 2016 1 commit
  29. 19 Dec, 2016 1 commit
  30. 18 Dec, 2016 1 commit
  31. 04 Nov, 2016 1 commit
  32. 02 Oct, 2016 1 commit
  33. 05 Sep, 2016 1 commit
  34. 23 Aug, 2016 1 commit