* Second and third positional arguments after the name of the
estimated parameter in the estimated_params block are only
considered in the optimization stage (not in the MCMC)
* Do not store bounds in bayestopt_, because bounds do not always
reflect restrictions implied by prior shapes.
* prior_bounds routine returns a structure (with fields lb and ub)
instead of a matrix.
the computation of the prior density.
bayestopt_.p1 is always the prior mean
bayestopt_.p2 is always the prior standard deviation
bayestopt_.p3 is unchanged
bayestopt_.p4 is unchanged
bayestopt_.p5 [new field] is the prior mode
bayestopt_.p6 [new field] is the first hyper-parameter of the prior density
bayestopt_.p7 [new field] is the second hyper-parameter of the prior density
These fields are defined in set_prior and are never changed after. In
the previous version of Dynare, the hyper parameters of the densities
were updated at each iteration of the optimization routine or the
metropolis.
Removed fields pmean and pstdev.
Vectorized the code in priordens.
Fixed the bug mentionned by Gianni. If a (logged) density is evaluated
outside the prior domain, the output of priordens if minus infinity
(instead of a complex number).
git-svn-id: https://www.dynare.org/svn/dynare/trunk@2556 ac1d8469-bf42-47a9-8791-bf33cf982152
* added preprocessor support for inverse gamma of type 2
* added support for this distribution in prior_bounds.m and rndprior.m
* other cosmetic changes
git-svn-id: https://www.dynare.org/svn/dynare/dynare_v4@1983 ac1d8469-bf42-47a9-8791-bf33cf982152
* replaced CDF, PDF, quantile functions of the gamma by a GPL'd one
* replaced quantile function of the chi-square by a GPL'd one
git-svn-id: https://www.dynare.org/svn/dynare/dynare_v4@1981 ac1d8469-bf42-47a9-8791-bf33cf982152
* replaced dnorm, pnorm and qnorm by GPL alternatives in distributions/
* only add distributions/ to the path if under Matlab without Statistics toolbox
git-svn-id: https://www.dynare.org/svn/dynare/dynare_v4@1979 ac1d8469-bf42-47a9-8791-bf33cf982152