stepan
/
dynare
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Efficiency & Cosmetic changes related to the design of bayestopt_ and 

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
time-shift
stepan 2009-04-06 14:38:37 +00:00
parent f02ea822ec
commit aa31417a05
23 changed files with 644 additions and 514 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
matlab/DsgeLikelihood.m
View File

@ -74,7 +74,7 @@ function [fval,cost_flag,ys,trend_coeff,info] = DsgeLikelihood(xparam1,gend,data
H(k,k) = xparam1(i+offset)*xparam1(i+offset); H(k,k) = xparam1(i+offset)*xparam1(i+offset);
end end
offset = offset+estim_params_.nvn; offset = offset+estim_params_.nvn;
end end
if estim_params_.ncx if estim_params_.ncx
for i=1:estim_params_.ncx for i=1:estim_params_.ncx
k1 =estim_params_.corrx(i,1); k1 =estim_params_.corrx(i,1);
@ -330,6 +330,6 @@ function [fval,cost_flag,ys,trend_coeff,info] = DsgeLikelihood(xparam1,gend,data
% ------------------------------------------------------------------------------ % ------------------------------------------------------------------------------
% Adds prior if necessary % Adds prior if necessary
% ------------------------------------------------------------------------------ % ------------------------------------------------------------------------------
lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p1,bayestopt_.p2,bayestopt_.p3,bayestopt_.p4); lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p6,bayestopt_.p7,bayestopt_.p3,bayestopt_.p4);
fval = (likelihood-lnprior); fval = (likelihood-lnprior);
options_.kalman_algo = kalman_algo; options_.kalman_algo = kalman_algo;

2
matlab/DsgeLikelihood_hh.m
View File

@ -322,7 +322,7 @@ function [fval,llik,cost_flag,ys,trend_coeff,info] = DsgeLikelihood_hh(xparam1,g
% ------------------------------------------------------------------------------ % ------------------------------------------------------------------------------
% Adds prior if necessary % Adds prior if necessary
% ------------------------------------------------------------------------------ % ------------------------------------------------------------------------------
lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p1,bayestopt_.p2,bayestopt_.p3,bayestopt_.p4); lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p6,bayestopt_.p7,bayestopt_.p3,bayestopt_.p4);
fval = (likelihood-lnprior); fval = (likelihood-lnprior);
options_.kalman_algo = kalman_algo; options_.kalman_algo = kalman_algo;
llik=[-lnprior; .5*lik(start:end)]; llik=[-lnprior; .5*lik(start:end)];

2
matlab/DsgeVarLikelihood.m
View File

@ -217,7 +217,7 @@ else
lik = .5*lik;% Minus likelihood lik = .5*lik;% Minus likelihood
end end
lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p1,bayestopt_.p2,bayestopt_.p3,bayestopt_.p4); lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p6,bayestopt_.p7,bayestopt_.p3,bayestopt_.p4);
fval = (lik-lnprior); fval = (lik-lnprior);
if (nargout == 6) if (nargout == 6)

42
matlab/GetPosteriorParametersStatistics.m
View File

@ -15,7 +15,7 @@ function oo_ = GetPosteriorParametersStatistics(estim_params_, M_, options_, bay
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% None. % None.
% Copyright (C) 2006-2008 Dynare Team % Copyright (C) 2006-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -96,14 +96,14 @@ if np
oo_ = Filloo(oo_,name,type,post_mean,hpd_interval,post_median,post_var,post_deciles,density); oo_ = Filloo(oo_,name,type,post_mean,hpd_interval,post_median,post_var,post_deciles,density);
end end
end end
disp(sprintf(pformat,name,bayestopt_.pmean(ip),... disp(sprintf(pformat,name,bayestopt_.p1(ip),...
post_mean, ... post_mean, ...
hpd_interval, ... hpd_interval, ...
pnames(bayestopt_.pshape(ip)+1,:), ... pnames(bayestopt_.pshape(ip)+1,:), ...
bayestopt_.pstdev(ip))); bayestopt_.p2(ip)));
if TeX if TeX
TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.pmean(ip),... TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),post_mean,sqrt(post_var),hpd_interval); bayestopt_.p2(ip),post_mean,sqrt(post_var),hpd_interval);
end end
ip = ip+1; ip = ip+1;
end end
@ -144,11 +144,11 @@ if nvx
M_.Sigma_e(k,k) = post_mean*post_mean; M_.Sigma_e(k,k) = post_mean*post_mean;
end end
end end
disp(sprintf(pformat,name,bayestopt_.pmean(ip),post_mean,hpd_interval,... disp(sprintf(pformat,name,bayestopt_.p1(ip),post_mean,hpd_interval,...
pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.pstdev(ip))); pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.p2(ip)));
if TeX if TeX
TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.pmean(ip),... TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),post_mean,sqrt(post_var),hpd_interval); bayestopt_.p2(ip),post_mean,sqrt(post_var),hpd_interval);
end end
ip = ip+1; ip = ip+1;
end end
@ -184,11 +184,11 @@ if nvn
oo_ = Filloo(oo_,name,type,post_mean,hpd_interval,post_median,post_var,post_deciles,density); oo_ = Filloo(oo_,name,type,post_mean,hpd_interval,post_median,post_var,post_deciles,density);
end end
end end
disp(sprintf(pformat,name,bayestopt_.pmean(ip),post_mean,hpd_interval, ... disp(sprintf(pformat,name,bayestopt_.p1(ip),post_mean,hpd_interval, ...
pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.pstdev(ip))); pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.p2(ip)));
if TeX if TeX
TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.pmean(ip),... TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),post_mean,sqrt(post_var),hpd_interval); bayestopt_.p2(ip),post_mean,sqrt(post_var),hpd_interval);
end end
ip = ip+1; ip = ip+1;
end end
@ -237,11 +237,11 @@ if ncx
M_.Sigma_e(k2,k1) = M_.Sigma_e(k1,k2); M_.Sigma_e(k2,k1) = M_.Sigma_e(k1,k2);
end end
end end
disp(sprintf(pformat, name,bayestopt_.pmean(ip),post_mean,hpd_interval, ... disp(sprintf(pformat, name,bayestopt_.p1(ip),post_mean,hpd_interval, ...
pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.pstdev(ip))); pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.p2(ip)));
if TeX if TeX
TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.pmean(ip),... TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),post_mean,sqrt(post_var),hpd_interval); bayestopt_.p2(ip),post_mean,sqrt(post_var),hpd_interval);
end end
ip = ip+1; ip = ip+1;
end end
@ -288,11 +288,11 @@ if ncn
post_median,post_var,post_deciles,density); post_median,post_var,post_deciles,density);
end end
end end
disp(sprintf(pformat, name,bayestopt_.pmean(ip),post_mean,hpd_interval, ... disp(sprintf(pformat, name,bayestopt_.p1(ip),post_mean,hpd_interval, ...
pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.pstdev(ip))); pnames(bayestopt_.pshape(ip)+1,:),bayestopt_.p2(ip)));
if TeX if TeX
TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.pmean(ip),... TeXCore(fid,name,deblank(pnames(bayestopt_.pshape(ip)+1,:)),bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),post_mean,sqrt(post_var),hpd_interval); bayestopt_.p2(ip),post_mean,sqrt(post_var),hpd_interval);
end end
ip = ip+1; ip = ip+1;
end end

87
matlab/distributions/compute_prior_mode.m Normal file
View File

@ -0,0 +1,87 @@
function m = compute_prior_mode(hyperparameters,shape)
% This function computes the mode of the prior distribution given the (two, three or four) hyperparameters
% of the prior distribution.
%
% INPUTS
% hyperparameters [double] 1*n vector of hyper parameters.
% shape [integer] scalar specifying the prior shape:
% shape=1 => Beta distribution,
% shape=2 => Gamma distribution,
% shape=3 => Gaussian distribution,
% shape=4 => Inverse Gamma (type 1) distribution,
% shape=5 => Uniform distribution,
% shape=6 => Inverse Gamma (type 2) distribution.
%
% OUTPUTS
% m [double] scalar or 2*1 vector, the prior mode.
%
% REMARKS
% [1] The size of the vector of hyperparameters is 3 when the Gamma or Inverse Gamma is shifted and 4 when
% the support of the Beta distribution is not [0,1].
% [2] The hyperparameters of the uniform distribution are the lower and upper bounds.
% [3] The uniform distribution has an infinity of modes. In this case the function returns the prior mean.
% [4] For the beta distribution we can have 1, 2 or an infinity of modes.
% Copyright (C) 2009 Dynare Team
%
% This file is part of Dynare.
%
% Dynare is free software: you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation, either version 3 of the License, or
% (at your option) any later version.
%
% Dynare is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details.
%
% You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>.
m = NaN ;
switch shape
case 1
if (hyperparameters(1)>1 && hyperparameters(2)>1)
m = (hyperparameters(1)-1)/(hyperparameters(1)+hyperparameters(2)-2) ;
elseif (hyperparameters(1)<1 && hyperparameters(2)<1)
m = [ 0 ; 1 ] ;
elseif ( hyperparameters(1)<1 && hyperparameters(2)>1-eps ) || ( abs(hyperparameters(1)-1)<2*eps && hyperparameters(2)>1 )
m = 0;
elseif ( hyperparameters(1)>1 && hyperparameters(2)<1+eps ) || ( abs(hyperparameters(1)-1)<2*eps && hyperparameters(2)<1 )
m = 1;
elseif ( abs(hyperparameters(1)-1)<2*eps && abs(hyperparameters(2)-1)<2*eps )% Uniform distribution!
m = .5 ;
end
if length(hyperparameters)==4
m = m*(hyperparameters(4)-hyperparameters(3)) + hyperparameters(3) ;
end
case 2
if hyperparameters(1)<1
m = 0;
else
m = (hyperparameters(1)-1)*hyperparameters(2);
end
if length(hyperparameters)>2
m = m + hyperparameters(3);
end
case 3
m = hyperparameters(1);
case 4
% s = hyperparameters(1)
% nu = hyperparameters(2)
m = sqrt((hyperparameters(2)-1)/hyperparameters(1));
if length(hyperparameters)>2
m = m + hyperparameters(3);
end
case 5
m = .5*(hyperparameters(2)-hyperparameters(1)) ;
case 6
% s = hyperparameters(1)
% nu = hyperparameters(2)
m = hyperparameters(1)/(hyperparameters(2)+2) ;
if length(hyperparameters)>2
m = m + hyperparameters(3) ;
end
otherwise
error('Unknown prior shape!')
end

84
matlab/draw_prior_density.m
View File

@ -1,6 +1,5 @@
function [x,f,abscissa,dens,binf,bsup] = draw_prior_density(indx,bayestopt_); function [x,f,abscissa,dens,binf,bsup] = draw_prior_density(indx,bayestopt_);
% function [x,f,abscissa,dens,binf,bsup] = draw_prior_density(indx) % Computes values of prior densities at many points (before plotting)
% Computes values of prior density at many points (before plotting)
% %
% INPUTS % INPUTS
% indx [integer] Parameter number. % indx [integer] Parameter number.
@ -13,9 +12,7 @@ function [x,f,abscissa,dens,binf,bsup] = draw_prior_density(indx,bayestopt_);
% dens [double] Row vector, density % dens [double] Row vector, density
% binf: [double] Scalar, first element of x % binf: [double] Scalar, first element of x
% bsup: [double] Scalar, last element of x % bsup: [double] Scalar, last element of x
%
% SPECIAL REQUIREMENTS
% none
% Copyright (C) 2004-2009 Dynare Team % Copyright (C) 2004-2009 Dynare Team
% %
@ -34,73 +31,58 @@ function [x,f,abscissa,dens,binf,bsup] = draw_prior_density(indx,bayestopt_);
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
pmean = bayestopt_.pmean; pshape = bayestopt_.pshape;
pshape = bayestopt_.pshape;
p1 = bayestopt_.p1;
p2 = bayestopt_.p2;
p3 = bayestopt_.p3; p3 = bayestopt_.p3;
p4 = bayestopt_.p4; p4 = bayestopt_.p4;
p6 = bayestopt_.p6;
p7 = bayestopt_.p7;
truncprior = 1e-3; truncprior = 1e-3;
steps = 200; steps = 200;
indx
switch pshape(indx) switch pshape(indx)
case 1 % Beta prior case 1% Beta prior
density = @(x,a,b,aa,bb) betapdf((x-aa)/(bb-aa), a, b)/(bb-aa); density = @(x,a,b,aa,bb) betapdf((x-aa)/(bb-aa), a, b)/(bb-aa);
mu = (p1(indx)-p3(indx))/(p4(indx)-p3(indx)); infbound = betainv(truncprior,p6(indx),p7(indx))*(p4(indx)-p3(indx))+p3(indx);
stdd = p2(indx)/(p4(indx)-p3(indx)); supbound = betainv(1-truncprior,p6(indx),p7(indx))*(p4(indx)-p3(indx))+p3(indx);
a = (1-mu)*mu^2/stdd^2 - mu;
b = a*(1/mu-1);
aa = p3(indx);
bb = p4(indx);
infbound = betainv(truncprior,a,b)*(bb-aa)+aa;
supbound = betainv(1-truncprior,a,b)*(bb-aa)+aa;
stepsize = (supbound-infbound)/steps; stepsize = (supbound-infbound)/steps;
abscissa = infbound:stepsize:supbound; abscissa = infbound:stepsize:supbound;
dens = density(abscissa,a,b,aa,bb); dens = density(abscissa,p6(indx),p7(indx),p3(indx),p4(indx));
case 2 % Generalized Gamma prior case 2% Generalized Gamma prior
density = @(x,a,b,c) gampdf(x-c,a,b); density = @(x,a,b,c) gampdf(x-c,a,b);
mu = p1(indx)-p3(indx); infbound = gaminv(truncprior,p6(indx),p7(indx))+p3(indx);
b = p2(indx)^2/mu; supbound = gaminv(1-truncprior,p6(indx),p7(indx))+p3(indx);
a = mu/b;
c = p3(indx);
infbound = gaminv(truncprior,a,b)+c;
supbound = gaminv(1-truncprior,a,b)+c;
stepsize = (supbound-infbound)/steps; stepsize = (supbound-infbound)/steps;
abscissa = infbound:stepsize:supbound; abscissa = infbound:stepsize:supbound;
dens = density(abscissa,a,b,c); dens = density(abscissa,p6(indx),p7(indx),p3(indx));
case 3 % Gaussian prior case 3% Gaussian prior
a = p1(indx); infbound = norminv(truncprior,p6(indx),p7(indx));
b = p2(indx); supbound = norminv(1-truncprior,p6(indx),p7(indx));
infbound = norminv(truncprior,a,b);
supbound = norminv(1-truncprior,a,b);
stepsize = (supbound-infbound)/steps; stepsize = (supbound-infbound)/steps;
abscissa = infbound:stepsize:supbound; abscissa = infbound:stepsize:supbound;
dens = normpdf(abscissa,a,b); dens = normpdf(abscissa,p6(indx),p7(indx));
case 4 % Inverse-gamma of type 1 prior case 4% Inverse-gamma of type 1 prior
nu = p2(indx); infbound = 1/sqrt(gaminv(1-10*truncprior, p7(indx)/2, 2/p6(indx)))+p3(indx);
s = p1(indx); supbound = 1/sqrt(gaminv(10*truncprior, p7(indx)/2, 2/p6(indx)))+p3(indx);
infbound = 1/sqrt(gaminv(1-10*truncprior, nu/2, 2/s));
supbound = 1/sqrt(gaminv(10*truncprior, nu/2, 2/s));
stepsize = (supbound-infbound)/steps; stepsize = (supbound-infbound)/steps;
abscissa = infbound:stepsize:supbound; abscissa = infbound:stepsize:supbound;
dens = exp(lpdfig1(abscissa,s,nu)); dens = exp(lpdfig1(abscissa-p3(indx),p6(indx),p7(indx)));
case 5 % Uniform prior case 5% Uniform prior
infbound = p1(indx); infbound = p6(indx);
supbound = p2(indx); supbound = p7(indx);
stepsize = (supbound-infbound)/steps; stepsize = (supbound-infbound)/steps;
abscissa = infbound:stepsize:supbound; abscissa = infbound:stepsize:supbound;
dens = ones(1, steps) / (supbound-infbound); dens = ones(1, steps) / (supbound-infbound);
case 6 % Inverse-gamma of type 2 prior case 6% Inverse-gamma of type 2 prior
nu = p2(indx); infbound = 1/(gaminv(1-10*truncprior, p7(indx)/2, 2/p6(indx)))+p3(indx);
s = p1(indx); supbound = 1/(gaminv(10*truncprior, p7(indx)/2, 2/p6(indx)))+p3(indx);
infbound = 1/(gaminv(1-10*truncprior, nu/2, 2/s)); stepsize = (supbound-infbound)/steps ;
supbound = 1/(gaminv(10*truncprior, nu/2, 2/s));
stepsize = (supbound-infbound)/steps;
abscissa = infbound:stepsize:supbound; abscissa = infbound:stepsize:supbound;
dens = exp(lpdfig2(abscissa,s,nu)); dens = exp(lpdfig2(abscissa-p3(indx),p6(indx),p7(indx)));
otherwise otherwise
error(sprintf('draw_prior_density: unknown distribution shape (index %d, type %d)', indx, pshape(indx))); error(sprintf('draw_prior_density: unknown distribution shape (index %d, type %d)', indx, pshape(indx)));
end end
k = [1:length(dens)]; k = [1:length(dens)];

2
matlab/dsge_posterior_kernel.m
View File

@ -314,6 +314,6 @@ function [fval,cost_flag,ys,trend_coeff,info] = DsgeLikelihood(xparam1,gend,data
% ------------------------------------------------------------------------------ % ------------------------------------------------------------------------------
% Adds prior if necessary % Adds prior if necessary
% ------------------------------------------------------------------------------ % ------------------------------------------------------------------------------
lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p1,bayestopt_.p2,bayestopt_.p3,bayestopt_.p4); lnprior = priordens(xparam1,bayestopt_.pshape,bayestopt_.p6,bayestopt_.p7,bayestopt_.p3,bayestopt_.p4);
fval = (likelihood-lnprior); fval = (likelihood-lnprior);
options_.kalman_algo = kalman_algo; options_.kalman_algo = kalman_algo;

49
matlab/dynare_estimation_1.m
View File

@ -118,6 +118,9 @@ else% If estim_params_ is empty...
bayestopt_.p2 = []; bayestopt_.p2 = [];
bayestopt_.p3 = []; bayestopt_.p3 = [];
bayestopt_.p4 = []; bayestopt_.p4 = [];
bayestopt_.p5 = [];
bayestopt_.p6 = [];
bayestopt_.p7 = [];
estim_params_.nvx = 0; estim_params_.nvx = 0;
estim_params_.nvn = 0; estim_params_.nvn = 0;
estim_params_.ncx = 0; estim_params_.ncx = 0;
@ -431,13 +434,13 @@ if options_.mode_compute > 0 & options_.posterior_mode_estimation
% covariance (a diagonal matrix) %%Except for infinite variances ;-) % covariance (a diagonal matrix) %%Except for infinite variances ;-)
varinit = 'prior'; varinit = 'prior';
if strcmpi(varinit,'prior') if strcmpi(varinit,'prior')
stdev = bayestopt_.pstdev; stdev = bayestopt_.p2;
indx = find(isinf(stdev)); indx = find(isinf(stdev));
stdev(indx) = ones(length(indx),1)*sqrt(10); stdev(indx) = ones(length(indx),1)*sqrt(10);
vars = stdev.^2; vars = stdev.^2;
CovJump = diag(vars); CovJump = diag(vars);
elseif strcmpi(varinit,'eye') elseif strcmpi(varinit,'eye')
vars = ones(length(bayestopt_.pstdev),1)*0.1; vars = ones(length(bayestopt_.p2),1)*0.1;
CovJump = diag(vars); CovJump = diag(vars);
else else
disp('gmhmaxlik :: Error!') disp('gmhmaxlik :: Error!')
@ -552,7 +555,7 @@ if options_.posterior_mode_estimation
invhess = inv(hh); invhess = inv(hh);
stdh = sqrt(diag(invhess)); stdh = sqrt(diag(invhess));
else else
variances = bayestopt_.pstdev.^2; variances = bayestopt_.p2.^2;
invhess = 0.01*diag(variances); invhess = 0.01*diag(variances);
%invhess = 0.01*eye(length(variances)); %invhess = 0.01*eye(length(variances));
end end
@ -575,9 +578,9 @@ if any(bayestopt_.pshape > 0) & options_.posterior_mode_estimation
name = bayestopt_.name{ip}; name = bayestopt_.name{ip};
disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', ... disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', ...
name, ... name, ...
bayestopt_.pmean(ip),xparam1(ip),stdh(ip),tstath(ip), ... bayestopt_.p1(ip),xparam1(ip),stdh(ip),tstath(ip), ...
pnames(bayestopt_.pshape(ip)+1,:), ... pnames(bayestopt_.pshape(ip)+1,:), ...
bayestopt_.pstdev(ip))); bayestopt_.p2(ip)));
eval(['oo_.posterior_mode.parameters.' name ' = xparam1(ip);']); eval(['oo_.posterior_mode.parameters.' name ' = xparam1(ip);']);
eval(['oo_.posterior_std.parameters.' name ' = stdh(ip);']); eval(['oo_.posterior_std.parameters.' name ' = stdh(ip);']);
ip = ip+1; ip = ip+1;
@ -591,9 +594,9 @@ if any(bayestopt_.pshape > 0) & options_.posterior_mode_estimation
k = estim_params_.var_exo(i,1); k = estim_params_.var_exo(i,1);
name = deblank(M_.exo_names(k,:)); name = deblank(M_.exo_names(k,:));
disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', ... disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', ...
name,bayestopt_.pmean(ip),xparam1(ip), ... name,bayestopt_.p1(ip),xparam1(ip), ...
stdh(ip),tstath(ip),pnames(bayestopt_.pshape(ip)+1,:), ... stdh(ip),tstath(ip),pnames(bayestopt_.pshape(ip)+1,:), ...
bayestopt_.pstdev(ip))); bayestopt_.p2(ip)));
M_.Sigma_e(k,k) = xparam1(ip)*xparam1(ip); M_.Sigma_e(k,k) = xparam1(ip)*xparam1(ip);
eval(['oo_.posterior_mode.shocks_std.' name ' = xparam1(ip);']); eval(['oo_.posterior_mode.shocks_std.' name ' = xparam1(ip);']);
eval(['oo_.posterior_std.shocks_std.' name ' = stdh(ip);']); eval(['oo_.posterior_std.shocks_std.' name ' = stdh(ip);']);
@ -607,10 +610,10 @@ if any(bayestopt_.pshape > 0) & options_.posterior_mode_estimation
for i=1:nvn for i=1:nvn
name = deblank(options_.varobs(estim_params_.var_endo(i,1),:)); name = deblank(options_.varobs(estim_params_.var_endo(i,1),:));
disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', ... disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', ...
name,bayestopt_.pmean(ip), ... name,bayestopt_.p1(ip), ...
xparam1(ip),stdh(ip),tstath(ip), ... xparam1(ip),stdh(ip),tstath(ip), ...
pnames(bayestopt_.pshape(ip)+1,:), ... pnames(bayestopt_.pshape(ip)+1,:), ...
bayestopt_.pstdev(ip))); bayestopt_.p2(ip)));
eval(['oo_.posterior_mode.measurement_errors_std.' name ' = xparam1(ip);']); eval(['oo_.posterior_mode.measurement_errors_std.' name ' = xparam1(ip);']);
eval(['oo_.posterior_std.measurement_errors_std.' name ' = stdh(ip);']); eval(['oo_.posterior_std.measurement_errors_std.' name ' = stdh(ip);']);
ip = ip+1; ip = ip+1;
@ -626,8 +629,8 @@ if any(bayestopt_.pshape > 0) & options_.posterior_mode_estimation
name = [deblank(M_.exo_names(k1,:)) ',' deblank(M_.exo_names(k2,:))]; name = [deblank(M_.exo_names(k1,:)) ',' deblank(M_.exo_names(k2,:))];
NAME = [deblank(M_.exo_names(k1,:)) '_' deblank(M_.exo_names(k2,:))]; NAME = [deblank(M_.exo_names(k1,:)) '_' deblank(M_.exo_names(k2,:))];
disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', name, ... disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', name, ...
bayestopt_.pmean(ip),xparam1(ip),stdh(ip),tstath(ip), ... bayestopt_.p1(ip),xparam1(ip),stdh(ip),tstath(ip), ...
pnames(bayestopt_.pshape(ip)+1,:), bayestopt_.pstdev(ip))); pnames(bayestopt_.pshape(ip)+1,:), bayestopt_.p2(ip)));
M_.Sigma_e(k1,k2) = xparam1(ip)*sqrt(M_.Sigma_e(k1,k1)*M_.Sigma_e(k2,k2)); M_.Sigma_e(k1,k2) = xparam1(ip)*sqrt(M_.Sigma_e(k1,k1)*M_.Sigma_e(k2,k2));
M_.Sigma_e(k2,k1) = M_.Sigma_e(k1,k2); M_.Sigma_e(k2,k1) = M_.Sigma_e(k1,k2);
eval(['oo_.posterior_mode.shocks_corr.' NAME ' = xparam1(ip);']); eval(['oo_.posterior_mode.shocks_corr.' NAME ' = xparam1(ip);']);
@ -645,8 +648,8 @@ if any(bayestopt_.pshape > 0) & options_.posterior_mode_estimation
name = [deblank(M_.endo_names(k1,:)) ',' deblank(M_.endo_names(k2,:))]; name = [deblank(M_.endo_names(k1,:)) ',' deblank(M_.endo_names(k2,:))];
NAME = [deblank(M_.endo_names(k1,:)) '_' deblank(M_.endo_names(k2,:))]; NAME = [deblank(M_.endo_names(k1,:)) '_' deblank(M_.endo_names(k2,:))];
disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', name, ... disp(sprintf('%12s %7.3f %8.4f %7.4f %7.4f %4s %6.4f', name, ...
bayestopt_.pmean(ip),xparam1(ip),stdh(ip),tstath(ip), ... bayestopt_.p1(ip),xparam1(ip),stdh(ip),tstath(ip), ...
pnames(bayestopt_.pshape(ip)+1,:), bayestopt_.pstdev(ip))); pnames(bayestopt_.pshape(ip)+1,:), bayestopt_.p2(ip)));
eval(['oo_.posterior_mode.measurement_errors_corr.' NAME ' = xparam1(ip);']); eval(['oo_.posterior_mode.measurement_errors_corr.' NAME ' = xparam1(ip);']);
eval(['oo_.posterior_std.measurement_errors_corr.' NAME ' = stdh(ip);']); eval(['oo_.posterior_std.measurement_errors_corr.' NAME ' = stdh(ip);']);
ip = ip+1; ip = ip+1;
@ -772,8 +775,8 @@ if any(bayestopt_.pshape > 0) & options_.TeX %% Bayesian estimation (posterior m
fprintf(fidTeX,'$%s$ & %s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n',... fprintf(fidTeX,'$%s$ & %s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n',...
M_.param_names_tex(estim_params_.param_vals(i,1),:),... M_.param_names_tex(estim_params_.param_vals(i,1),:),...
deblank(pnames(bayestopt_.pshape(ip)+1,:)),... deblank(pnames(bayestopt_.pshape(ip)+1,:)),...
bayestopt_.pmean(ip),... bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),... bayestopt_.p2(ip),...
xparam1(ip),... xparam1(ip),...
stdh(ip)); stdh(ip));
ip = ip + 1; ip = ip + 1;
@ -808,8 +811,8 @@ if any(bayestopt_.pshape > 0) & options_.TeX %% Bayesian estimation (posterior m
fprintf(fidTeX,[ '$%s$ & %4s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n'],... fprintf(fidTeX,[ '$%s$ & %4s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n'],...
deblank(M_.exo_names_tex(k,:)),... deblank(M_.exo_names_tex(k,:)),...
deblank(pnames(bayestopt_.pshape(ip)+1,:)),... deblank(pnames(bayestopt_.pshape(ip)+1,:)),...
bayestopt_.pmean(ip),... bayestopt_.p1(ip),...
bayestopt_.pstdev(ip),... bayestopt_.p2(ip),...
xparam1(ip), ... xparam1(ip), ...
stdh(ip)); stdh(ip));
ip = ip+1; ip = ip+1;
@ -843,8 +846,8 @@ if any(bayestopt_.pshape > 0) & options_.TeX %% Bayesian estimation (posterior m
fprintf(fidTeX,'$%s$ & %4s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n',... fprintf(fidTeX,'$%s$ & %4s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n',...
deblank(M_.endo_names_tex(idx,:)), ... deblank(M_.endo_names_tex(idx,:)), ...
deblank(pnames(bayestopt_.pshape(ip)+1,:)), ... deblank(pnames(bayestopt_.pshape(ip)+1,:)), ...
bayestopt_.pmean(ip), ... bayestopt_.p1(ip), ...
bayestopt_.pstdev(ip),... bayestopt_.p2(ip),...
xparam1(ip),... xparam1(ip),...
stdh(ip)); stdh(ip));
ip = ip+1; ip = ip+1;
@ -878,8 +881,8 @@ if any(bayestopt_.pshape > 0) & options_.TeX %% Bayesian estimation (posterior m
fprintf(fidTeX,[ '$%s$ & %s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n'],... fprintf(fidTeX,[ '$%s$ & %s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n'],...
[deblank(M_.exo_names_tex(k1,:)) ',' deblank(M_.exo_names_tex(k2,:))], ... [deblank(M_.exo_names_tex(k1,:)) ',' deblank(M_.exo_names_tex(k2,:))], ...
deblank(pnames(bayestopt_.pshape(ip)+1,:)), ... deblank(pnames(bayestopt_.pshape(ip)+1,:)), ...
bayestopt_.pmean(ip), ... bayestopt_.p1(ip), ...
bayestopt_.pstdev(ip), ... bayestopt_.p2(ip), ...
xparam1(ip), ... xparam1(ip), ...
stdh(ip)); stdh(ip));
ip = ip+1; ip = ip+1;
@ -913,8 +916,8 @@ if any(bayestopt_.pshape > 0) & options_.TeX %% Bayesian estimation (posterior m
fprintf(fidTeX,'$%s$ & %s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n',... fprintf(fidTeX,'$%s$ & %s & %7.3f & %6.4f & %8.4f & %7.4f \\\\ \n',...
[deblank(M_.endo_names_tex(k1,:)) ',' deblank(M_.endo_names_tex(k2,:))], ... [deblank(M_.endo_names_tex(k1,:)) ',' deblank(M_.endo_names_tex(k2,:))], ...
pnames(bayestopt_.pshape(ip)+1,:), ... pnames(bayestopt_.pshape(ip)+1,:), ...
bayestopt_.pmean(ip), ... bayestopt_.p1(ip), ...
bayestopt_.pstdev(ip), ... bayestopt_.p2(ip), ...
xparam1(ip), ... xparam1(ip), ...
stdh(ip)); stdh(ip));
ip = ip+1; ip = ip+1;

2
matlab/forcst_unc.m
View File

@ -123,7 +123,7 @@ function forcst_unc(y0,var_list)
end end
% compute shock uncertainty around forecast with mean prior % compute shock uncertainty around forecast with mean prior
set_parameters(bayestopt_.pmean); set_parameters(bayestopt_.p1);
[dr,info] = resol(oo_.steady_state,0); [dr,info] = resol(oo_.steady_state,0);
[yf3,yf3_intv] = forcst(dr,y0,periods,var_list); [yf3,yf3_intv] = forcst(dr,y0,periods,var_list);
yf3_1 = yf3'-[zeros(maximum_lag,n); yf3_intv]; yf3_1 = yf3'-[zeros(maximum_lag,n); yf3_intv];

7
matlab/get_prior_info.m
View File

@ -54,8 +54,8 @@ function get_prior_info(info)
for i=1:size(bayestopt_.name,1) for i=1:size(bayestopt_.name,1)
[tmp,TexName] = get_the_name(i,1); [tmp,TexName] = get_the_name(i,1);
PriorShape = PriorNames{ bayestopt_.pshape(i) }; PriorShape = PriorNames{ bayestopt_.pshape(i) };
PriorMean = bayestopt_.pmean(i); PriorMean = bayestopt_.p1(i);
PriorStandardDeviation = bayestopt_.pstdev(i); PriorStandardDeviation = bayestopt_.p2(i);
switch bayestopt_.pshape(i) switch bayestopt_.pshape(i)
case { 1 , 5 } case { 1 , 5 }
LowerBound = bayestopt_.p3(i); LowerBound = bayestopt_.p3(i);
@ -102,10 +102,9 @@ function get_prior_info(info)
function format_string = build_format_string(bayestopt,i) function format_string = build_format_string(bayestopt,i)
format_string = ['%s & %s & %6.4f &']; format_string = ['%s & %s & %6.4f &'];
if isinf(bayestopt.pstdev(i)) if isinf(bayestopt.p2(i))
format_string = [ format_string , ' %s &']; format_string = [ format_string , ' %s &'];
else else
format_string = [ format_string , ' %6.4f &']; format_string = [ format_string , ' %6.4f &'];

32
matlab/lpdfgam.m
View File

@ -1,20 +1,19 @@
function ldens = lpdfgam(x,a,b); function ldens = lpdfgam(x,a,b);
% Evaluates the logged GAMMA PDF at x.
% function ldens = lpdfgam(x,a,b) %
% log GAMMA PDF % INPUTS
% x [double] m*n matrix of locations,
% a [double] m*n matrix or scalar, First GAMMA distribution parameters,
% b [double] m*n matrix or scalar, Second GAMMA distribution parameters,
% %
% INPUTS
% x: density evatuated at x
% a: GAMMA distribution paramerer
% b: GAMMA distribution paramerer
%
% OUTPUTS % OUTPUTS
% ldens: the log GAMMA PDF % ldens [double] m*n matrix of logged GAMMA densities evaluated at x.
%
% %
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% none % none
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -31,8 +30,11 @@ function ldens = lpdfgam(x,a,b);
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
ldens = -gammaln(a) -a*log(b)+ (a-1)*log(x) -x/b ; ldens = -Inf( size(x) ) ;
idx = find( x>0 );
% 10/11/03 MJ adapted from an earlier GAUSS version by F. Schorfeide,
% translated to MATLAB by R. Wouters if length(a)==1
% use MATLAB gammaln rather than lngam ldens(idx) = -gammaln(a) - a*log(b) + (a-1)*log(x(idx)) - x(idx)/b ;
else
ldens(idx) = -gammaln(a(idx)) - a(idx).*log(b(idx)) + (a(idx)-1).*log(x(idx)) - x(idx)./b(idx) ;
end

38
matlab/lpdfgbeta.m
View File

@ -1,22 +1,20 @@
function ldens = lpdfgbeta(x,a,b,aa,bb); function ldens = lpdfgbeta(x,a,b,aa,bb);
% Evaluates the logged BETA PDF at x.
% function ldens = lpdfgbeta(x,a,b,aa,bb);
% log (generalized) BETA PDF
% %
% INPUTS % INPUTS
% x: density evatuated at x % x [double] m*n matrix of loactions,
% a: BETA distribution paramerer % a [double] m*n matrix of First BETA distribution parameters,
% b: BETA distribution paramerer % b [double] m*n matrix of Second BETA distribution parameters,
% aa: lower bound % aa [double] m*n matrix of lower bounds,
% bb: upper bound % bb [double] m*n matrix of upper bounds.
%
% OUTPUTS % OUTPUTS
% ldens: the log (generalized) BETA PDF % ldens [double] m*n matrix of logged (generalized) BETA densities.
% %
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% none % none
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -33,9 +31,11 @@ function ldens = lpdfgbeta(x,a,b,aa,bb);
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
ldens = -betaln(a,b) + (a-1)*log(x-aa) + (b-1)*log(bb-x) - (a+b-1)*log(bb-aa); ldens = -Inf( size(x) ) ;
idx = find( (x-aa)>0 & (x-bb)<0 ) ;
%gammaln(a+b) - gammaln(a) - gammaln(b)
%betaln(a,b) if length(a)==1
%pause ldens(idx) = -betaln(a,b) + (a-1)*log(x(idx)-aa) + (b-1)*log(bb-x(idx)) - (a+b-1)*log(bb-aa) ;
% 02/16/04 SA Interval [aa,bb] is the support of the probability density function. else
ldens(idx) = -betaln(a(idx),b(idx)) + (a(idx)-1).*log(x(idx)-aa(idx)) + (b(idx)-1).*log(bb(idx)-x(idx)) - (a(idx)+b(idx)-1).*log(bb(idx)-aa(idx));
end

36
matlab/lpdfig1.m
View File

@ -1,22 +1,23 @@
function ldens = lpdfig1(x,s,nu) function ldens = lpdfig1(x,s,nu)
% function ldens = lpdfig1(x,s,nu) % Evaluates the logged INVERSE-GAMMA-1 PDF at x.
% log INVERSE GAMMA (type 1)
% X ~ IG1(s,nu)
% X = sqrt(Y) where Y ~ IG2(s,nu) and Y = inv(Z) with Z ~ G(nu/2,2/s) (Gamma distribution)
% %
% INPUTS % X ~ IG1(s,nu) if X = sqrt(Y) where Y ~ IG2(s,nu) and Y = inv(Z) with Z ~ G(nu/2,2/s) (Gamma distribution)
% x: density evatuated at x %
% s: shape parameter % See L. Bauwens, M. Lubrano and J-F. Richard [1999, appendix A] for more details.
% nu: scale parameter %
%
% INPUTS
% x [double] m*n matrix of locations,
% s [double] m*n matrix or scalar, First INVERSE-GAMMA-1 distribution parameters,
% nu [double] m*n matrix or scalar, Second INVERSE-GAMMA-1 distribution parameters.
% %
% OUTPUTS % OUTPUTS
% ldens: the log INVERSE GAMMA density function (type 1) % ldens [double] m*n matrix of logged INVERSE-GAMMA-1 densities evaluated at x.
% %
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% See L. Bauwens, M. Lubrano and J-F. Richard [1999, appendix A] for more % none
% details.
% Copyright (C) 2004-2008 Dynare Team % Copyright (C) 2004-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -33,4 +34,11 @@ function ldens = lpdfig1(x,s,nu)
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
ldens = log(2) - gammaln(nu/2) - (nu/2).*log(2/s) - (nu+1)*log(x) - .5*s./(x.^2); ldens = -Inf( size(x) ) ;
idx = find( x>0 ) ;
if length(s)==1
ldens(idx) = log(2) - gammaln(.5*nu) - .5*nu*(log(2)-log(s)) - (nu+1)*log(x(idx)) - .5*s./(x(idx).*x(idx)) ;
else
ldens(idx) = log(2) - gammaln(.5*nu(idx)) - .5*nu(idx).*(log(2)-log(s(idx))) - (nu(idx)+1).*log(x(idx)) - .5*s(idx)./(x(idx).*x(idx)) ;
end

36
matlab/lpdfig2.m
View File

@ -1,22 +1,23 @@
function ldens = lpdfig2(x,s,nu) function ldens = lpdfig2(x,s,nu)
% function ldens = lpdfig2(x,s,nu) % Evaluates the logged INVERSE-GAMMA-2 PDF at x.
% log INVERSE GAMMA (type 2)
% X ~ IG2(s,nu)
% X = inv(Z) where Z ~ G(nu/2,2/s) (Gamma distribution)
% %
% INPUTS % X ~ IG2(s,nu) if X = inv(Z) where Z ~ G(nu/2,2/s) (Gamma distribution)
% x: density evatuated at x %
% s: shape parameter % See L. Bauwens, M. Lubrano and J-F. Richard [1999, appendix A] for more details.
% nu: scale parameter %
%
% INPUTS
% x [double] m*n matrix of locations,
% s [double] m*n matrix or scalar, First INVERSE-GAMMA-2 distribution parameters,
% nu [double] m*n matrix or scalar, Second INVERSE-GAMMA-2 distribution parameters.
% %
% OUTPUTS % OUTPUTS
% ldens: the log INVERSE GAMMA density function (type 2) % ldens [double] m*n matrix of logged INVERSE-GAMMA-2 densities evaluated at x.
% %
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% See L. Bauwens, M. Lubrano and J-F. Richard [1999, appendix A] for more % none
% details.
% Copyright (C) 2004-2008 Dynare Team % Copyright (C) 2004-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -33,4 +34,11 @@ function ldens = lpdfig2(x,s,nu)
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
ldens = - gammaln(nu/2) - (nu/2)*log(2/s) - .5*(nu+2)*log(x) -.5*s./x; ldens = -Inf( size(x) ) ;
idx = find( x>0 ) ;
if length(s)==1
ldens(idx) = -gammaln(.5*nu) - (.5*nu)*(log(2)-log(s)) - .5*(nu+2)*log(x(idx)) -.5*s./x(idx);
else
ldens(idx) = -gammaln(.5*nu(idx)) - (.5*nu(idx)).*(log(2)-log(s(idx))) - .5*(nu(idx)+2).*log(x(idx)) -.5*s(idx)./x(idx);
end

27
matlab/lpdfnorm.m
View File

@ -1,19 +1,19 @@
function f = lpdfnorm(x,m,s) function ldens = lpdfnorm(x,a,b)
% function f = lpdfnorm(x,m,s) % Evaluates the logged UNIVARIATE GAUSSIAN PDF at x.
% The log of the normal density function
% %
% INPUTS % INPUTS
% x: density evatuated at x % x [double] m*n matrix of locations,
% m: mean % a [double] m*n matrix or scalar, First GAUSSIAN distribution parameters (expectation)
% s: standard deviation % b [double] m*n matrix or scalar, Second GAUSSIAN distribution parameters (standard deviation).
% %
% OUTPUTS % OUTPUTS
% f: the log of the normal density function % ldens [double] m*n matrix of logged GAUSSIAN densities evaluated at x.
%
% %
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% none % none
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -30,7 +30,6 @@ function f = lpdfnorm(x,m,s)
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
if nargin<3, s=1; end if nargin<3, b=1; end
if nargin<2, m=0; end if nargin<2, a=0; end
f = -log(s)-log(2*pi)/2-((x-m)./s).^2/2; ldens = -log(b) -.5*log(2*pi) - .5*((x-a)./b).*((x-a)./b) ;

4
matlab/plot_priors.m
View File

@ -13,7 +13,7 @@ function plot_priors(bayestopt_,M_,options_)
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% None % None
% Copyright (C) 2004-2008 Dynare Team % Copyright (C) 2004-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -33,7 +33,7 @@ function plot_priors(bayestopt_,M_,options_)
TeX = options_.TeX; TeX = options_.TeX;
figurename = 'Priors'; figurename = 'Priors';
npar = length(bayestopt_.pmean); npar = length(bayestopt_.p1);
[nbplt,nr,nc,lr,lc,nstar] = pltorg(npar); [nbplt,nr,nc,lr,lc,nstar] = pltorg(npar);
if TeX if TeX

42
matlab/prior_bounds.m
View File

@ -11,7 +11,7 @@ function bounds = prior_bounds(bayestopt)
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% none % none
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -31,41 +31,33 @@ function bounds = prior_bounds(bayestopt)
global options_ global options_
pshape = bayestopt.pshape; pshape = bayestopt.pshape;
pmean = bayestopt.pmean;
p1 = bayestopt.p1;
p2 = bayestopt.p2;
p3 = bayestopt.p3; p3 = bayestopt.p3;
p4 = bayestopt.p4; p4 = bayestopt.p4;
p6 = bayestopt.p6;
p7 = bayestopt.p7;
n = length(pmean); bounds = zeros(length(p6),2);
bounds = zeros(n,2);
for i=1:n for i=1:length(p6)
switch pshape(i) switch pshape(i)
case 1 case 1
mu = (pmean(i)-p3(i))/(p4(i)-p3(i)); bounds(i,1) = betainv(options_.prior_trunc,p6(i),p7(i))*(p4(i)-p3(i))+p3(i);
stdd = p2(i)/(p4(i)-p3(i)); bounds(i,2) = betainv(1-options_.prior_trunc,p6(i),p7(i))*(p4(i)-p3(i))+p3(i);
A = (1-mu)*mu^2/stdd^2 - mu;
B = A*(1/mu - 1);
bounds(i,1) = betainv(options_.prior_trunc,A,B)*(p4(i)-p3(i))+p3(i);
bounds(i,2) = betainv(1-options_.prior_trunc,A,B)*(p4(i)-p3(i))+p3(i);
case 2 case 2
b = p2(i)^2/(pmean(i)-p3(i)); bounds(i,1) = gaminv(options_.prior_trunc,p6(i),p7(i))+p3(i);
a = (pmean(i)-p3(i))/b; bounds(i,2) = gaminv(1-options_.prior_trunc,p6(i),p7(i))+p3(i);
bounds(i,1) = gaminv(options_.prior_trunc,a,b)+p3(i);
bounds(i,2) = gaminv(1-options_.prior_trunc,a,b)+p3(i);
case 3 case 3
bounds(i,1) = norminv(options_.prior_trunc,pmean(i),p2(i)); bounds(i,1) = norminv(options_.prior_trunc,p6(i),p7(i));
bounds(i,2) = norminv(1-options_.prior_trunc,pmean(i),p2(i)); bounds(i,2) = norminv(1-options_.prior_trunc,p6(i),p7(i));
case 4 case 4
bounds(i,1) = 1/sqrt(gaminv(1-options_.prior_trunc, p2(i)/2, 2/p1(i))); bounds(i,1) = 1/sqrt(gaminv(1-options_.prior_trunc, p7(i)/2, 2/p6(i)))+p3(i);
bounds(i,2) = 1/sqrt(gaminv(options_.prior_trunc, p2(i)/2, 2/p1(i))); bounds(i,2) = 1/sqrt(gaminv(options_.prior_trunc, p7(i)/2, 2/p6(i)))+p3(i);
case 5 case 5
bounds(i,1) = p1(i)+(p2(i)-p1(i))*options_.prior_trunc; bounds(i,1) = p6(i)+(p7(i)-p6(i))*options_.prior_trunc;
bounds(i,2) = p2(i)-(p2(i)-p1(i))*options_.prior_trunc; bounds(i,2) = p7(i)-(p7(i)-p6(i))*options_.prior_trunc;
case 6 case 6
bounds(i,1) = 1/gaminv(1-options_.prior_trunc, p2(i)/2, 2/p1(i)); bounds(i,1) = 1/gaminv(1-options_.prior_trunc, p7(i)/2, 2/p6(i))+p3(i);
bounds(i,2) = 1/gaminv(options_.prior_trunc, p2(i)/2, 2/p1(i)); bounds(i,2) = 1/gaminv(options_.prior_trunc, p7(i)/2, 2/p6(i))+p3(i);
otherwise otherwise
error(sprintf('prior_bounds: unknown distribution shape (index %d, type %d)', i, pshape(i))); error(sprintf('prior_bounds: unknown distribution shape (index %d, type %d)', i, pshape(i)));
end end

82
matlab/prior_draw.m
View File

@ -32,63 +32,73 @@ function pdraw = prior_draw(init, prior_structure)
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
persistent prior_mean prior_standard_deviation a b p1 p2 p3 p4 persistent p6 p7 p3 p4
persistent uniform_index gaussian_index gamma_index beta_index inverse_gamma_1_index inverse_gamma_2_index persistent uniform_index gaussian_index gamma_index beta_index inverse_gamma_1_index inverse_gamma_2_index
persistent uniform_draws gaussian_draws gamma_draws beta_draws inverse_gamma_1_draws inverse_gamma_2_draws
if nargin>0 && init if nargin>0 && init
prior_shape = prior_structure.pshape; p6 = prior_structure.p6;
prior_mean = prior_structure.pmean; p7 = prior_structure.p7;
prior_standard_deviation = prior_structure.pstdev;
p1 = prior_structure.p1;
p2 = prior_structure.p2;
p3 = prior_structure.p3; p3 = prior_structure.p3;
p4 = prior_structure.p4; p4 = prior_structure.p4;
number_of_estimated_parameters = length(p1); number_of_estimated_parameters = length(p6);
a = NaN(number_of_estimated_parameters,1);
b = NaN(number_of_estimated_parameters,1);
beta_index = find(prior_shape==1); beta_index = find(prior_shape==1);
beta_draws = 1;
if isempty(beta_index)
beta_draws = 0;
end
gamma_index = find(prior_shape==2); gamma_index = find(prior_shape==2);
gamma_draws = 1;
if isempty(gamma_index)
gamma_draws = 0;
end
gaussian_index = find(prior_shape==3); gaussian_index = find(prior_shape==3);
gaussian_draws = 1;
if isempty(gaussian_index)
gaussian_draws = 0;
end
inverse_gamma_1_index = find(prior_shape==4); inverse_gamma_1_index = find(prior_shape==4);
inverse_gamma_1_draws = 1;
if isempty(inverse_gamma_1_index)
inverse_gamma_1_draws = 0;
end
uniform_index = find(prior_shape==5); uniform_index = find(prior_shape==5);
uniform_draws = 1;
if isempty(uniform_index)
uniform_draws = 0;
end
inverse_gamma_2_index = find(prior_shape==6); inverse_gamma_2_index = find(prior_shape==6);
% Set parameters for the beta prior inverse_gamma_2_draws = 1;
mu = (p1(beta_index)-p3(beta_index))./(p4(beta_index)-p3(beta_index)); if isempty(inverse_gamma_2_index)
stdd = p2(beta_index)./(p4(beta_index)-p3(beta_index)); inverse_gamma_2_draws = 0;
a(beta_index) = (1-mu).*mu.^2./stdd.^2 - mu; end
b(beta_index) = a(beta_index).*(1./mu - 1);
% Set parameters for the gamma prior
mu = p1(gamma_index)-p3(gamma_index);
b(gamma_index) = p2(gamma_index).^2./mu;
a(gamma_index) = mu./b(gamma_index);
% Initialization of the vector of prior draws.
pdraw = zeros(number_of_estimated_parameters,1); pdraw = zeros(number_of_estimated_parameters,1);
return return
end end
% Uniform draws. if uniform_draws
if ~isempty(uniform_index)
pdraw(uniform_index) = rand(length(uniform_index),1).*(p4(uniform_index)-p3(uniform_index)) + p3(uniform_index); pdraw(uniform_index) = rand(length(uniform_index),1).*(p4(uniform_index)-p3(uniform_index)) + p3(uniform_index);
end end
% Gaussian draws.
if ~isempty(gaussian_index) if gaussian_draws
pdraw(gaussian_index) = randn(length(gaussian_index),1).*prior_standard_deviation(gaussian_index) + prior_mean(gaussian_index); pdraw(gaussian_index) = randn(length(gaussian_index),1).*p7(gaussian_index) + p6(gaussian_index);
end end
% Gamma draws.
if ~isempty(gamma_index) if gamma_draws
pdraw(gamma_index) = gamrnd(a(gamma_index),b(gamma_index))+p3(gamma_index); pdraw(gamma_index) = gamrnd(p6(gamma_index),p7(gamma_index))+p3(gamma_index);
end end
% Beta draws.
if ~isempty(beta_index) if beta_draws
pdraw(beta_index) = (p4(beta_index)-p3(beta_index)).*betarnd(a(beta_index),b(beta_index))+p3(beta_index); pdraw(beta_index) = (p4(beta_index)-p3(beta_index)).*betarnd(p6(beta_index),p7(beta_index))+p3(beta_index);
end end
% Inverted gamma (type 1) draws.
if ~isempty(inverse_gamma_1_index) if inverse_gamma_1_draws
pdraw(inverse_gamma_1_index) = ... pdraw(inverse_gamma_1_index) = ...
sqrt(1./gamrnd(p2(inverse_gamma_1_index)/2,2./p1(inverse_gamma_1_index))); sqrt(1./gamrnd(p7(inverse_gamma_1_index)/2,2./p6(inverse_gamma_1_index)))+p3(inverse_gamma_1_index);
end end
% Inverted gamma (type 2) draws.
if ~isempty(inverse_gamma_2_index) if inverse_gamma_2_draws
pdraw(inverse_gamma_2_index) = ... pdraw(inverse_gamma_2_index) = ...
1./gamrnd(p2(inverse_gamma_2_index)/2,2./p1(inverse_gamma_2_index)); 1./gamrnd(p7(inverse_gamma_2_index)/2,2./p6(inverse_gamma_2_index))+p3(inverse_gamma_2_index);
end end

2
matlab/prior_sampler.m
View File

@ -45,7 +45,7 @@ function results = prior_sampler(drsave,M_,bayestopt_,options_,oo_)
count_dll_problem = 0; count_dll_problem = 0;
count_unknown_problem = 0; count_unknown_problem = 0;
NumberOfSimulations = options_.prior_mc; NumberOfSimulations = options_.prior_mc;
NumberOfParameters = length(bayestopt_.pmean); NumberOfParameters = length(bayestopt_.p1);
NumberOfEndogenousVariables = size(M_.endo_names,1); NumberOfEndogenousVariables = size(M_.endo_names,1);
NumberOfElementsPerFile = ceil(options_.MaxNumberOfBytes/NumberOfParameters/NumberOfEndogenousVariables/8) NumberOfElementsPerFile = ceil(options_.MaxNumberOfBytes/NumberOfParameters/NumberOfEndogenousVariables/8)
if NumberOfSimulations <= NumberOfElementsPerFile if NumberOfSimulations <= NumberOfElementsPerFile

152
matlab/priordens.m
View File

@ -1,29 +1,19 @@
function lnprior = priordens(para, pshape, p1, p2, p3, p4) function logged_prior_density = priordens(x, pshape, p6, p7, p3, p4)
% Computes a prior density for the structural parameters of DSGE models
% function lnprior = priordens(para, pshape, p1, p2, p3, p4)
% computes a prior density for the structural parameters of DSGE models
% %
% INPUTS % INPUTS
% para: parameter value % x [double] vector with n elements.
% pshape: 0 is point mass, both para and p2 are ignored % pshape [integer] vector with n elements (bayestopt_.pshape).
% 1 is BETA % p6: [double] vector with n elements, first parameter of the prior distribution (bayestopt_.p6).
% 2 is GAMMA % p7: [double] vector with n elements, second parameter of the prior distribution (bayestopt_.p7).
% 3 is NORMAL % p3: [double] vector with n elements, lower bounds.
% 4 is INVERTED GAMMA TYPE I % p4: [double] vector with n elements, upper bound.
% 5 is UNIFORM
% 6 is INVERTED GAMMA TYPE II
% p1: mean
% p2: standard deviation
% p3: lower bound
% p4: upper bound
% %
% OUTPUTS % OUTPUTS
% lnprior: log of the prior density % logged_prior_density [double] scalar, log of the prior density evaluated at x.
% %
% SPECIAL REQUIREMENTS
% none
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -40,39 +30,89 @@ function lnprior = priordens(para, pshape, p1, p2, p3, p4)
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
lnprior = 0; persistent pflag
nprio = length(pshape); persistent id1 id2 id3 id4 id5 id6
persistent tt1 tt2 tt3 tt4 tt5 tt6
i = 1; if isempty(pflag)
while i <= nprio; Number0fParameters = length(pshape);
a = 0; % Beta indices.
b = 0; tt1 = 1;
if pshape(i) == 1; % (generalized) BETA Prior id1 = find(pshape==1);
mu = (p1(i)-p3(i))/(p4(i)-p3(i)); if isempty(id1)
stdd = p2(i)/(p4(i)-p3(i)); tt1 = 0;
a = (1-mu)*mu^2/stdd^2 - mu; end
b = a*(1/mu - 1); % Gamma indices.
lnprior = lnprior + lpdfgbeta(para(i),a,b,p3(i),p4(i)) ; tt2 = 1;
elseif pshape(i) == 2; % GAMMA PRIOR id2 = find(pshape==2);
b = p2(i)^2/(p1(i)-p3(i)); if isempty(id2)
a = (p1(i)-p3(i))/b; tt2 = 0;
lnprior = lnprior + lpdfgam(para(i)-p3(i),a,b); end
elseif pshape(i) == 3; % GAUSSIAN PRIOR % Gaussian indices.
lnprior = lnprior + lpdfnorm(para(i),p1(i),p2(i)); tt3 = 1;
elseif pshape(i) == 4; % INVGAMMA1 PRIOR id3 = find(pshape==3);
lnprior = lnprior + lpdfig1(para(i),p1(i),p2(i)); if isempty(id3)
elseif pshape(i) == 5; % UNIFORM PRIOR tt3 = 0;
lnprior = lnprior + log(1/(p2(i)-p1(i))); end
elseif pshape(i) == 6; % INVGAMMA2 PRIOR % Inverse-Gamma-1 indices.
lnprior = lnprior + lpdfig2(para(i),p1(i),p2(i)); tt4 = 1;
end; id4 = find(pshape==4);
i = i+1; if isempty(id4)
end; tt4 = 0;
end
% Uniform indices.
tt5 = 1;
id5 = find(pshape==5);
if isempty(id5)
tt5 = 0;
end
% Inverse-Gamma-2 indices.
tt6 = 1;
id6 = find(pshape==6);
if isempty(id6)
tt6 = 0;
end
pflag = 1;
end
% 10/11/03 MJ adapted from an earlier version in GAUSS by F. Schorfheide logged_prior_density = 0.0;
% and translated to Matlab by R. Wouters
% 11/18/03 MJ adopted M.Ratto's suggestion for inverse gamma if tt1
% changed order of input parameters logged_prior_density = logged_prior_density + sum(lpdfgbeta(x(id1),p6(id1),p7(id1),p3(id1),p4(id1))) ;
% 01/16/04 MJ added invgamma2 if isinf(logged_prior_density)
% for invgamma p2 is now standard error return
% 16/02/04 SA changed beta prior call end
end
if tt2
logged_prior_density = logged_prior_density + sum(lpdfgam(x(id2)-p3(id2),p6(id2),p7(id2))) ;
if isinf(logged_prior_density)
return
end
end
if tt3
logged_prior_density = logged_prior_density + sum(lpdfnorm(x(id3),p6(id3),p7(id3))) ;
end
if tt4
logged_prior_density = logged_prior_density + sum(lpdfig1(x(id4)-p3(id4),p6(id4),p7(id4))) ;
if isinf(logged_prior_density)
return
end
end
if tt5
if any(x(id5)-p3(id5)<0) || any(x(id5)-p4(id5)>0)
logged_prior_density = -Inf ;
return
end
logged_prior_density = logged_prior_density + sum(log(1./(p4(id5)-p3(id5)))) ;
end
if tt6
logged_prior_density = logged_prior_density + sum(lpdfig2(x(id6)-p3(id6),p6(id6),p7(id6))) ;
if isinf(logged_prior_density)
return
end
end

69
matlab/rndprior.m
View File

@ -11,7 +11,7 @@ function y = rndprior(bayestopt_)
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% none % none
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -29,52 +29,29 @@ function y = rndprior(bayestopt_)
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
pshape=bayestopt_.pshape; pshape=bayestopt_.pshape;
pmean=bayestopt_.pmean;
p1=bayestopt_.p1;
p2=bayestopt_.p2;
p3=bayestopt_.p3; p3=bayestopt_.p3;
p4=bayestopt_.p4; p4=bayestopt_.p4;
p6=bayestopt_.p6;
for i=1:length(pmean), p7=bayestopt_.p7;
y = NaN(1,length(pshape));
for i=1:length(pshape)
switch pshape(i) switch pshape(i)
case 1 % Beta
case 1 % Beta y(i) = betarnd(p6(i),p7(i));
mu = (pmean(i)-p3(i))/(p4(i)-p3(i)); y(i) = y(1,i) * (p4(i)-p3(i)) + p3(i);
stdd = p2(i)/(p4(i)-p3(i)); case 2 % Generalized gamma
A = (1-mu)*mu^2/stdd^2 - mu; y(i) = gamrnd(p6(i),p7(i)) + p3(i);
B = A*(1/mu - 1); case 3 % Gaussian
y(1,i) = betarnd(A, B); y(i) = randn*p7(i) + p6(i) ;
y(1,i) = y(1,i) * (p4(i)-p3(i)) + p3(i); case 4 % Inverse-gamma type 1
y(i) = 1/sqrt(gamrnd(p7(i)/2, 2/p6(i))) + p3(i);
case 2 % Generalized gamma case 5 % Uniform
mu = pmean(i)-p3(i); y(i) = rand*(p4(i)-p3(i)) + p3(i);
B = p2(i)^2/mu; case 6 % Inverse-gamma type 2
A = mu/B; y(i) = 1/gamrnd(p7(i)/2, 2/p6(i)) + p3(i);
y(1,i) = gamrnd(A, B) + p3(i); otherwise
error(sprintf('rndprior: unknown distribution shape (index %d, type %d)', i, pshape(i)));
case 3 % Gaussian
MU = pmean(i);
SIGMA = p2(i);
y(1,i) = randn*SIGMA+ MU;
case 4 % Inverse-gamma type 1
nu = p2(i);
s = p1(i);
y(1,i) = 1/sqrt(gamrnd(nu/2, 2/s));
case 5 % Uniform
y(1,i) = rand*(p2(i)-p1(i)) + p1(i);
case 6 % Inverse-gamma type 2
nu = p2(i);
s = p1(i);
y(1,i) = 1/gamrnd(nu/2, 2/s);
otherwise
error(sprintf('rndprior: unknown distribution shape (index %d, type %d)', i, pshape(i)));
end end
end end
% initial version by Marco Ratto

337
matlab/set_prior.m
View File

@ -18,7 +18,7 @@ function [xparam1, estim_params_, bayestopt_, lb, ub, M_]=set_prior(estim_params
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% None % None
% Copyright (C) 2003-2008 Dynare Team % Copyright (C) 2003-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -35,165 +35,190 @@ function [xparam1, estim_params_, bayestopt_, lb, ub, M_]=set_prior(estim_params
% You should have received a copy of the GNU General Public License % You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
nvx = size(estim_params_.var_exo,1); nvx = size(estim_params_.var_exo,1);
nvn = size(estim_params_.var_endo,1); nvn = size(estim_params_.var_endo,1);
ncx = size(estim_params_.corrx,1); ncx = size(estim_params_.corrx,1);
ncn = size(estim_params_.corrn,1); ncn = size(estim_params_.corrn,1);
np = size(estim_params_.param_vals,1); np = size(estim_params_.param_vals,1);
estim_params_.nvx = nvx;
estim_params_.nvn = nvn;
estim_params_.ncx = ncx;
estim_params_.ncn = ncn;
estim_params_.np = np;
estim_params_.nvx = nvx; xparam1 = [];
estim_params_.nvn = nvn; ub = [];
estim_params_.ncx = ncx; lb = [];
estim_params_.ncn = ncn; bayestopt_.pshape = [];
estim_params_.np = np; bayestopt_.p1 = []; % prior mean
bayestopt_.p2 = []; % prior standard deviation
xparam1 = []; bayestopt_.p3 = []; % lower bound
ub = []; bayestopt_.p4 = []; % upper bound
lb = []; bayestopt_.p5 = []; % prior mode
bayestopt_.pshape = []; bayestopt_.p6 = []; % first hyper-parameter (\alpha for the BETA and GAMMA distributions, s for the INVERSE GAMMAs, expectation for the GAUSSIAN distribution, lower bound for the UNIFORM distribution).
bayestopt_.pmean = []; bayestopt_.p7 = []; % second hyper-parameter (\beta for the BETA and GAMMA distributions, \nu for the INVERSE GAMMAs, standard deviation for the GAUSSIAN distribution, upper bound for the UNIFORM distribution).
bayestopt_.pstdev = [];
bayestopt_.p1 = []; bayestopt_.jscale = [];
bayestopt_.p2 = []; bayestopt_.name = [];
bayestopt_.p3 = []; if nvx
bayestopt_.p4 = []; xparam1 = estim_params_.var_exo(:,2);
bayestopt_.jscale = []; ub = estim_params_.var_exo(:,4);
bayestopt_.name = []; lb = estim_params_.var_exo(:,3);
if nvx bayestopt_.pshape = estim_params_.var_exo(:,5);
xparam1 = estim_params_.var_exo(:,2); bayestopt_.p1 = estim_params_.var_exo(:,6);
ub = estim_params_.var_exo(:,4); bayestopt_.p2 = estim_params_.var_exo(:,7);
lb = estim_params_.var_exo(:,3); bayestopt_.p3 = estim_params_.var_exo(:,8);
bayestopt_.pshape = estim_params_.var_exo(:,5); bayestopt_.p4 = estim_params_.var_exo(:,9);
bayestopt_.pmean = estim_params_.var_exo(:,6); bayestopt_.jscale = estim_params_.var_exo(:,10);
bayestopt_.pstdev = estim_params_.var_exo(:,7); bayestopt_.name = cellstr(M_.exo_names(estim_params_.var_exo(:,1),:));
bayestopt_.p3 = estim_params_.var_exo(:,8);
bayestopt_.p4 = estim_params_.var_exo(:,9);
bayestopt_.jscale = estim_params_.var_exo(:,10);
bayestopt_.name = cellstr(M_.exo_names(estim_params_.var_exo(:,1),:));
end
if nvn
if M_.H == 0
nvarobs = size(options_.varobs,1);
M_.H = zeros(nvarobs,nvarobs);
end end
for i=1:nvn if nvn
estim_params_.var_endo(i,1) = strmatch(deblank(M_.endo_names(estim_params_.var_endo(i,1),:)),deblank(options_.varobs),'exact'); if M_.H == 0
nvarobs = size(options_.varobs,1);
M_.H = zeros(nvarobs,nvarobs);
end
for i=1:nvn
estim_params_.var_endo(i,1) = strmatch(deblank(M_.endo_names(estim_params_.var_endo(i,1),:)),deblank(options_.varobs),'exact');
end
xparam1 = [xparam1; estim_params_.var_endo(:,2)];
ub = [ub; estim_params_.var_endo(:,4)];
lb = [lb; estim_params_.var_endo(:,3)];
bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.var_endo(:,5)];
bayestopt_.p1 = [ bayestopt_.p1; estim_params_.var_endo(:,6)];
bayestopt_.p2 = [ bayestopt_.p2; estim_params_.var_endo(:,7)];
bayestopt_.p3 = [ bayestopt_.p3; estim_params_.var_endo(:,8)];
bayestopt_.p4 = [ bayestopt_.p4; estim_params_.var_endo(:,9)];
bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.var_endo(:,10)];
bayestopt_.name = cellstr(strvcat(char(bayestopt_.name), M_.endo_names(estim_params_.var_endo(:,1),:)));
end end
xparam1 = [xparam1; estim_params_.var_endo(:,2)]; if ncx
ub = [ub; estim_params_.var_endo(:,4)]; xparam1 = [xparam1; estim_params_.corrx(:,3)];
lb = [lb; estim_params_.var_endo(:,3)]; ub = [ub; max(min(estim_params_.corrx(:,5),1),-1)];
bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.var_endo(:,5)]; lb = [lb; max(min(estim_params_.corrx(:,4),1),-1)];
bayestopt_.pmean = [ bayestopt_.pmean; estim_params_.var_endo(:,6)]; bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.corrx(:,6)];
bayestopt_.pstdev = [ bayestopt_.pstdev; estim_params_.var_endo(:,7)]; bayestopt_.p1 = [ bayestopt_.p1; estim_params_.corrx(:,7)];
bayestopt_.p3 = [ bayestopt_.p3; estim_params_.var_endo(:,8)]; bayestopt_.p2 = [ bayestopt_.p2; estim_params_.corrx(:,8)];
bayestopt_.p4 = [ bayestopt_.p4; estim_params_.var_endo(:,9)]; bayestopt_.p3 = [ bayestopt_.p3; estim_params_.corrx(:,9)];
bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.var_endo(:,10)]; bayestopt_.p4 = [ bayestopt_.p4; estim_params_.corrx(:,10)];
bayestopt_.name = cellstr(strvcat(char(bayestopt_.name),... bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.corrx(:,11)];
M_.endo_names(estim_params_.var_endo(:,1),:))); bayestopt_.name = cellstr(strvcat(char(bayestopt_.name), char(strcat(cellstr(M_.exo_names(estim_params_.corrx(:,1),:)), ...
end ',' , cellstr(M_.exo_names(estim_params_.corrx(:,2),:))))));
if ncx end
xparam1 = [xparam1; estim_params_.corrx(:,3)]; if ncn
ub = [ub; max(min(estim_params_.corrx(:,5),1),-1)]; if M_.H == 0
lb = [lb; max(min(estim_params_.corrx(:,4),1),-1)]; nvarobs = size(options_.varobs,1);
bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.corrx(:,6)]; M_.H = zeros(nvarobs,nvarobs);
bayestopt_.pmean = [ bayestopt_.pmean; estim_params_.corrx(:,7)]; end
bayestopt_.pstdev = [ bayestopt_.pstdev; estim_params_.corrx(:,8)]; xparam1 = [xparam1; estim_params_.corrn(:,3)];
bayestopt_.p3 = [ bayestopt_.p3; estim_params_.corrx(:,9)]; ub = [ub; max(min(estim_params_.corrn(:,5),1),-1)];
bayestopt_.p4 = [ bayestopt_.p4; estim_params_.corrx(:,10)]; lb = [lb; max(min(estim_params_.corrn(:,4),1),-1)];
bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.corrx(:,11)]; bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.corrn(:,6)];
bayestopt_.name = cellstr(strvcat(char(bayestopt_.name),... bayestopt_.p1 = [ bayestopt_.p1; estim_params_.corrn(:,7)];
char(strcat(cellstr(M_.exo_names(estim_params_.corrx(:,1),:)),... bayestopt_.p2 = [ bayestopt_.p2; estim_params_.corrn(:,8)];
',',... bayestopt_.p3 = [ bayestopt_.p3; estim_params_.corrn(:,9)];
cellstr(M_.exo_names(estim_params_.corrx(:,2),:)))))); bayestopt_.p4 = [ bayestopt_.p4; estim_params_.corrn(:,10)];
end bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.corrn(:,11)];
if ncn bayestopt_.name = cellstr(strvcat(char(bayestopt_.name), char(strcat(cellstr(M_.endo_names(estim_params_.corrn(:,1),:)),...
if M_.H == 0 ',' , cellstr(M_.endo_names(estim_params_.corrn(:,2),:))))));
nvarobs = size(options_.varobs,1); end
M_.H = zeros(nvarobs,nvarobs); if np
xparam1 = [xparam1; estim_params_.param_vals(:,2)];
ub = [ub; estim_params_.param_vals(:,4)];
lb = [lb; estim_params_.param_vals(:,3)];
bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.param_vals(:,5)];
bayestopt_.p1 = [ bayestopt_.p1; estim_params_.param_vals(:,6)];
bayestopt_.p2 = [ bayestopt_.p2; estim_params_.param_vals(:,7)];
bayestopt_.p3 = [ bayestopt_.p3; estim_params_.param_vals(:,8)];
bayestopt_.p4 = [ bayestopt_.p4; estim_params_.param_vals(:,9)];
bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.param_vals(:,10)];
bayestopt_.name = cellstr(strvcat(char(bayestopt_.name),M_.param_names(estim_params_.param_vals(:,1),:)));
end end
xparam1 = [xparam1; estim_params_.corrn(:,3)];
ub = [ub; max(min(estim_params_.corrn(:,5),1),-1)];
lb = [lb; max(min(estim_params_.corrn(:,4),1),-1)];
bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.corrn(:,6)];
bayestopt_.pmean = [ bayestopt_.pmean; estim_params_.corrn(:,7)];
bayestopt_.pstdev = [ bayestopt_.pstdev; estim_params_.corrn(:,8)];
bayestopt_.p3 = [ bayestopt_.p3; estim_params_.corrn(:,9)];
bayestopt_.p4 = [ bayestopt_.p4; estim_params_.corrn(:,10)];
bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.corrn(:,11)];
bayestopt_.name = cellstr(strvcat(char(bayestopt_.name),...
char(strcat(cellstr(M_.endo_names(estim_params_.corrn(:,1),:)),...
',',...
cellstr(M_.endo_names(estim_params_.corrn(:,2),:))))));
end
if np
xparam1 = [xparam1; estim_params_.param_vals(:,2)];
ub = [ub; estim_params_.param_vals(:,4)];
lb = [lb; estim_params_.param_vals(:,3)];
bayestopt_.pshape = [ bayestopt_.pshape; estim_params_.param_vals(:,5)];
bayestopt_.pmean = [ bayestopt_.pmean; estim_params_.param_vals(:,6)];
bayestopt_.pstdev = [ bayestopt_.pstdev; estim_params_.param_vals(:,7)];
bayestopt_.p3 = [ bayestopt_.p3; estim_params_.param_vals(:,8)];
bayestopt_.p4 = [ bayestopt_.p4; estim_params_.param_vals(:,9)];
bayestopt_.jscale = [ bayestopt_.jscale; estim_params_.param_vals(:, ...
10)];
bayestopt_.name = cellstr(strvcat(char(bayestopt_.name),M_.param_names(estim_params_.param_vals(:,1),:)));
end
bayestopt_.ub = ub; bayestopt_.ub = ub;
bayestopt_.lb = lb; bayestopt_.lb = lb;
bayestopt_.p6 = NaN(size(bayestopt_.p1)) ;
bayestopt_.p7 = bayestopt_.p6 ;
bayestopt_.p1 = bayestopt_.pmean; % generalized location parameters by default for beta distribution
bayestopt_.p2 = bayestopt_.pstdev; k = find(bayestopt_.pshape == 1);
k1 = find(isnan(bayestopt_.p3(k)));
% generalized location parameters by default for beta distribution bayestopt_.p3(k(k1)) = zeros(length(k1),1);
k = find(bayestopt_.pshape == 1); k1 = find(isnan(bayestopt_.p4(k)));
k1 = find(isnan(bayestopt_.p3(k))); bayestopt_.p4(k(k1)) = ones(length(k1),1);
bayestopt_.p3(k(k1)) = zeros(length(k1),1); for i=1:length(k)
k1 = find(isnan(bayestopt_.p4(k))); mu = (bayestopt_.p1(k(i))-bayestopt_.p3(k(i)))/(bayestopt_.p4(k(i))-bayestopt_.p3(k(i)));
bayestopt_.p4(k(k1)) = ones(length(k1),1); stdd = bayestopt_.p2(k(i))/(bayestopt_.p4(k(i))-bayestopt_.p3(k(i)));
bayestopt_.p6(k(i)) = (1-mu)*mu^2/stdd^2 - mu ;
% generalized location parameter by default for gamma distribution bayestopt_.p7(k(i)) = bayestopt_.p6(k(i))*(1/mu-1) ;
k = find(bayestopt_.pshape == 2); m = compute_prior_mode([ bayestopt_.p6(k(i)) , bayestopt_.p7(k(i)) ],1);
k1 = find(isnan(bayestopt_.p3(k))); if length(m)==1
k2 = find(isnan(bayestopt_.p4(k))); bayestopt_.p5(k(i)) = m;
bayestopt_.p3(k(k1)) = zeros(length(k1),1); else
bayestopt_.p4(k(k2)) = Inf(length(k2),1); disp(['Prior distribution for parameter ' int2str(k(i)) ' has two modes!'])
bayestopt_.p5(k(i)) = bayestopt_.p1(k(i)) ;
% truncation parameters by default for normal distribution end
k = find(bayestopt_.pshape == 3); end
k1 = find(isnan(bayestopt_.p3(k)));
bayestopt_.p3(k(k1)) = -Inf*ones(length(k1),1); % generalized location parameter by default for gamma distribution
k1 = find(isnan(bayestopt_.p4(k))); k = find(bayestopt_.pshape == 2);
bayestopt_.p4(k(k1)) = Inf*ones(length(k1),1); k1 = find(isnan(bayestopt_.p3(k)));
k2 = find(isnan(bayestopt_.p4(k)));
bayestopt_.p3(k(k1)) = zeros(length(k1),1);
bayestopt_.p4(k(k2)) = Inf(length(k2),1);
for i=1:length(k)
mu = bayestopt_.p1(k(i))-bayestopt_.p3(k(i));
bayestopt_.p7(k(i)) = bayestopt_.p2(k(i))^2/mu ;
bayestopt_.p6(k(i)) = mu/bayestopt_.p7(k(i)) ;
bayestopt_.p5(k(i)) = compute_prior_mode([ bayestopt_.p6(k(i)) , bayestopt_.p7(k(i)) ], 2) ;
end
% truncation parameters by default for normal distribution
k = find(bayestopt_.pshape == 3);
k1 = find(isnan(bayestopt_.p3(k)));
bayestopt_.p3(k(k1)) = -Inf*ones(length(k1),1);
k1 = find(isnan(bayestopt_.p4(k)));
bayestopt_.p4(k(k1)) = Inf*ones(length(k1),1);
for i=1:length(k)
bayestopt_.p6(k(i)) = bayestopt_.p1(k(i)) ;
bayestopt_.p7(k(i)) = bayestopt_.p2(k(i)) ;
bayestopt_.p5(k(i)) = bayestopt_.p1(k(i)) ;
end
% inverse gamma distribution % inverse gamma distribution
k = find(bayestopt_.pshape == 4); k = find(bayestopt_.pshape == 4);
for i=1:length(k) k1 = find(isnan(bayestopt_.p3(k)));
[bayestopt_.p1(k(i)),bayestopt_.p2(k(i))] = ... k2 = find(isnan(bayestopt_.p4(k)));
inverse_gamma_specification(bayestopt_.pmean(k(i)),bayestopt_.pstdev(k(i)),1); bayestopt_.p3(k(k1)) = zeros(length(k1),1);
end bayestopt_.p4(k(k2)) = Inf(length(k2),1);
k1 = find(isnan(bayestopt_.p3(k))); for i=1:length(k)
k2 = find(isnan(bayestopt_.p4(k))); [bayestopt_.p6(k(i)),bayestopt_.p7(k(i))] = ...
bayestopt_.p3(k(k1)) = zeros(length(k1),1); inverse_gamma_specification(bayestopt_.p1(k(i))-bayestopt_.p3(k(i)),bayestopt_.p2(k(i)),1) ;
bayestopt_.p4(k(k2)) = Inf(length(k2),1); bayestopt_.p5(k(i)) = compute_prior_mode([ bayestopt_.p6(k(i)) , bayestopt_.p7(k(i)) , bayestopt_.p3(k(i)) ], 4) ;
end
% uniform distribution
k = find(bayestopt_.pshape == 5); % uniform distribution
for i=1:length(k) k = find(bayestopt_.pshape == 5);
[bayestopt_.pmean(k(i)),bayestopt_.pstdev(k(i)),bayestopt_.p1(k(i)),bayestopt_.p2(k(i))] = ... for i=1:length(k)
uniform_specification(bayestopt_.pmean(k(i)),bayestopt_.pstdev(k(i)),bayestopt_.p3(k(i)),bayestopt_.p4(k(i))); [bayestopt_.p1(k(i)),bayestopt_.p2(k(i)),bayestopt_.p6(k(i)),bayestopt_.p7(k(i))] = ...
end uniform_specification(bayestopt_.p1(k(i)),bayestopt_.p2(k(i)),bayestopt_.p3(k(i)),bayestopt_.p4(k(i)));
bayestopt_.p3(k(i)) = bayestopt_.p6(k(i)) ;
% inverse gamma distribution (type 2) bayestopt_.p4(k(i)) = bayestopt_.p7(k(i)) ;
k = find(bayestopt_.pshape == 6); bayestopt_.p5(k(i)) = NaN ;
for i=1:length(k) end
[bayestopt_.p1(k(i)),bayestopt_.p2(k(i))] = ...
inverse_gamma_specification(bayestopt_.pmean(k(i)),bayestopt_.pstdev(k(i)),2); % inverse gamma distribution (type 2)
end k = find(bayestopt_.pshape == 6);
k1 = find(isnan(bayestopt_.p3(k))); k1 = find(isnan(bayestopt_.p3(k)));
k2 = find(isnan(bayestopt_.p4(k))); k2 = find(isnan(bayestopt_.p4(k)));
bayestopt_.p3(k(k1)) = zeros(length(k1),1); bayestopt_.p3(k(k1)) = zeros(length(k1),1);
bayestopt_.p4(k(k2)) = Inf(length(k2),1); bayestopt_.p4(k(k2)) = Inf(length(k2),1);
for i=1:length(k)
k = find(isnan(xparam1)); [bayestopt_.p6(k(i)),bayestopt_.p7(k(i))] = ...
xparam1(k) = bayestopt_.pmean(k); inverse_gamma_specification(bayestopt_.p1(k(i))-bayestopt_.p3(k(i)),bayestopt_.p2(k(i)),2);
bayestopt_.p5(k(i)) = compute_prior_mode([ bayestopt_.p6(k(i)) , bayestopt_.p7(k(i)) , bayestopt_.p3(k(i)) ], 6) ;
end
k = find(isnan(xparam1));
xparam1(k) = bayestopt_.p1(k);

20
matlab/uniform_specification.m
View File

@ -1,6 +1,4 @@
function [m,s,p1,p2] = uniform_specification(m,s,p3,p4) function [m,s,p6,p7] = uniform_specification(m,s,p3,p4)
% function [m,s,p1,p2] = uniform_specification(m,s,p3,p4)
% Specification of the uniform density function parameters % Specification of the uniform density function parameters
% %
% INPUTS % INPUTS
@ -18,7 +16,7 @@ function [m,s,p1,p2] = uniform_specification(m,s,p3,p4)
% SPECIAL REQUIREMENTS % SPECIAL REQUIREMENTS
% none % none
% Copyright (C) 2004-2008 Dynare Team % Copyright (C) 2004-2009 Dynare Team
% %
% This file is part of Dynare. % This file is part of Dynare.
% %
@ -36,11 +34,11 @@ function [m,s,p1,p2] = uniform_specification(m,s,p3,p4)
% along with Dynare. If not, see <http://www.gnu.org/licenses/>. % along with Dynare. If not, see <http://www.gnu.org/licenses/>.
if ~(isnan(p3) | isnan(p4)) if ~(isnan(p3) | isnan(p4))
p1 = p3; p6 = p3;
p2 = p4; p7 = p4;
m = (p3+p4)/2; m = (p3+p4)/2;
s = (p4-p3)/(sqrt(12)); s = (p4-p3)/(sqrt(12));
else else
p1 = m-s*sqrt(3); p6 = m-s*sqrt(3);
p2 = m+s*sqrt(3); p7 = m+s*sqrt(3);
end end