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

Added new option (diffuse_kalman_tol) and fixed tolerance paremeters in diffuse smoother routines.

time-shift
Stéphane Adjemian (Charybdis) 2015-04-03 17:48:25 +02:00
parent e97e5c3407
commit 090c4fedbd
7 changed files with 25 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
doc/dynare.texi
View File

@ -5221,6 +5221,8 @@ singularity is encountered, Dynare by default automatically switches to the univ
@item kalman_tol = @var{DOUBLE}
@anchor{kalman_tol} Numerical tolerance for determining the singularity of the covariance matrix of the prediction errors during the Kalman filter (minimum allowed reciprocal of the matrix condition number). Default value is @code{1e-10}
@item diffuse_kalman_tol = @var{DOUBLE}
@anchor{diffuse_kalman_tol} Numerical tolerance for determining the singularity of the covariance matrix of the prediction errors (@math{F_{\infty}}) and the rank of the covariance matrix of the state variables (@math{P_{\infty}}) during the Diffuse Kalman filter. Default value is @code{1e-8}
@item filter_covariance
@anchor{filter_covariance} Saves the series of one step ahead error of

5
matlab/DsgeSmoother.m
View File

@ -175,6 +175,7 @@ elseif options_.lik_init == 5 % Old diffuse Kalman filter only for th
Pinf = [];
end
kalman_tol = options_.kalman_tol;
diffuse_kalman_tol = options_.diffuse_kalman_tol;
riccati_tol = options_.riccati_tol;
data1 = Y-trend;
% -----------------------------------------------------------------------------
@ -200,7 +201,7 @@ if kalman_algo == 1 || kalman_algo == 3
[alphahat,epsilonhat,etahat,ahat,P,aK,PK,decomp] = missing_DiffuseKalmanSmootherH1_Z(ST, ...
Z,R1,Q,H,Pinf,Pstar, ...
data1,vobs,np,smpl,data_index, ...
options_.nk,kalman_tol,options_.filter_decomposition);
options_.nk,kalman_tol,diffuse_kalman_tol,options_.filter_decomposition);
if isinf(alphahat)
if kalman_algo == 1
kalman_algo = 2;
@ -227,7 +228,7 @@ if kalman_algo == 2 || kalman_algo == 4
[alphahat,epsilonhat,etahat,ahat,P,aK,PK,decomp] = missing_DiffuseKalmanSmootherH3_Z(ST, ...
Z,R1,Q,diag(H), ...
Pinf,Pstar,data1,vobs,np,smpl,data_index, ...
options_.nk,kalman_tol,...
options_.nk,kalman_tol,diffuse_kalman_tol, ...
options_.filter_decomposition);
end

1
matlab/global_initialization.m
View File

@ -387,6 +387,7 @@ options_.nobs = NaN;
options_.kalman_algo = 0;
options_.fast_kalman = 0;
options_.kalman_tol = 1e-10;
options_.diffuse_kalman_tol = 1e-8;
options_.use_univariate_filters_if_singularity_is_detected = 1;
options_.riccati_tol = 1e-6;
options_.lik_algo = 1;

16
matlab/missing_DiffuseKalmanSmootherH1_Z.m
View File

@ -1,6 +1,6 @@
function [alphahat,epsilonhat,etahat,atilde,P,aK,PK,decomp] = missing_DiffuseKalmanSmootherH1_Z(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,decomp_flag)
function [alphahat,epsilonhat,etahat,atilde,P,aK,PK,decomp] = missing_DiffuseKalmanSmootherH1_Z(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,diffuse_kalman_tol,decomp_flag)
% function [alphahat,epsilonhat,etahat,a,aK,PK,decomp] = DiffuseKalmanSmoother1(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,decomp_flag)
% function [alphahat,epsilonhat,etahat,a,aK,PK,decomp] = DiffuseKalmanSmoother1(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,diffuse_kalman_tol,decomp_flag)
% Computes the diffuse kalman smoother without measurement error, in the case of a non-singular var-cov matrix
%
% INPUTS
@ -18,6 +18,7 @@ function [alphahat,epsilonhat,etahat,atilde,P,aK,PK,decomp] = missing_DiffuseKal
% data_index [cell] 1*smpl cell of column vectors of indices.
% nk number of forecasting periods
% kalman_tol tolerance for reciprocal condition number
% diffuse_kalman_tol tolerance for reciprocal condition number (for Finf) and the rank of Pinf
% decomp_flag if true, compute filter decomposition
%
% OUTPUTS
@ -38,7 +39,7 @@ function [alphahat,epsilonhat,etahat,atilde,P,aK,PK,decomp] = missing_DiffuseKal
% Models", S.J. Koopman and J. Durbin (2003, in Journal of Time Series
% Analysis, vol. 24(1), pp. 85-98).
% Copyright (C) 2004-2011 Dynare Team
% Copyright (C) 2004-2015 Dynare Team
%
% This file is part of Dynare.
%
@ -80,7 +81,6 @@ Kstar = zeros(mm,pp,smpl);
P = zeros(mm,mm,smpl+1);
Pstar = zeros(spstar(1),spstar(2),smpl+1); Pstar(:,:,1) = Pstar1;
Pinf = zeros(spinf(1),spinf(2),smpl+1); Pinf(:,:,1) = Pinf1;
crit1 = 1.e-8;
steady = smpl;
rr = size(Q,1);
QQ = R*Q*transpose(R);
@ -91,7 +91,7 @@ epsilonhat = zeros(rr,smpl);
r = zeros(mm,smpl+1);
t = 0;
while rank(Pinf(:,:,t+1),crit1) && t<smpl
while rank(Pinf(:,:,t+1),diffuse_kalman_tol) && t<smpl
t = t+1;
di = data_index{t};
if isempty(di)
@ -103,8 +103,8 @@ while rank(Pinf(:,:,t+1),crit1) && t<smpl
ZZ = Z(di,:);
v(di,t)= Y(di,t) - ZZ*a(:,t);
Finf = ZZ*Pinf(:,:,t)*ZZ';
if rcond(Finf) < kalman_tol
if ~all(abs(Finf(:)) < kalman_tol)
if rcond(Finf) < diffuse_kalman_tol
if ~all(abs(Finf(:)) < diffuse_kalman_tol)
% The univariate diffuse kalman filter should be used.
alphahat = Inf;
return
@ -170,7 +170,7 @@ while notsteady && t<smpl
F = ZZ*P(:,:,t)*ZZ' + H(di,di);
if rcond(F) < kalman_tol
alphahat = Inf;
return
return
end
iF(di,di,t) = inv(F);
PZI = P(:,:,t)*ZZ'*iF(di,di,t);

14
matlab/missing_DiffuseKalmanSmootherH3_Z.m
View File

@ -1,4 +1,4 @@
function [alphahat,epsilonhat,etahat,a,P,aK,PK,decomp] = missing_DiffuseKalmanSmootherH3_Z(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,decomp_flag)
function [alphahat,epsilonhat,etahat,a,P,aK,PK,decomp] = missing_DiffuseKalmanSmootherH3_Z(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,diffuse_kalman_tol,decomp_flag)
% function [alphahat,epsilonhat,etahat,a1,P,aK,PK,d,decomp] = missing_DiffuseKalmanSmootherH3_Z(T,Z,R,Q,H,Pinf1,Pstar1,Y,pp,mm,smpl,data_index,nk,kalman_tol,decomp_flag)
% Computes the diffuse kalman smoother without measurement error, in the case of a singular var-cov matrix.
% Univariate treatment of multivariate time series.
@ -17,7 +17,8 @@ function [alphahat,epsilonhat,etahat,a,P,aK,PK,decomp] = missing_DiffuseKalmanSm
% smpl: sample size
% data_index [cell] 1*smpl cell of column vectors of indices.
% nk number of forecasting periods
% kalman_tol tolerance for zero divider
% kalman_tol tolerance for zero divider
% diffuse_kalman_tol tolerance for zero divider
% decomp_flag if true, compute filter decomposition
%
% OUTPUTS
@ -38,7 +39,7 @@ function [alphahat,epsilonhat,etahat,a,P,aK,PK,decomp] = missing_DiffuseKalmanSm
% Models", S.J. Koopman and J. Durbin (2003, in Journal of Time Series
% Analysis, vol. 24(1), pp. 85-98).
% Copyright (C) 2004-2011 Dynare Team
% Copyright (C) 2004-2015 Dynare Team
%
% This file is part of Dynare.
%
@ -80,7 +81,6 @@ Pstar = zeros(spstar(1),spstar(2),smpl); Pstar(:,:,1) = Pstar1;
Pinf = zeros(spinf(1),spinf(2),smpl); Pinf(:,:,1) = Pinf1;
Pstar1 = Pstar;
Pinf1 = Pinf;
crit1 = 1.e-6;
steady = smpl;
rr = size(Q,1); % number of structural shocks
QQ = R*Q*transpose(R);
@ -92,7 +92,7 @@ r = zeros(mm,smpl);
t = 0;
icc=0;
newRank = rank(Pinf(:,:,1),crit1);
newRank = rank(Pinf(:,:,1),diffuse_kalman_tol);
while newRank && t < smpl
t = t+1;
a(:,t) = a1(:,t);
@ -121,7 +121,7 @@ while newRank && t < smpl
end
end
if newRank
oldRank = rank(Pinf(:,:,t),crit1);
oldRank = rank(Pinf(:,:,t),diffuse_kalman_tol);
else
oldRank = 0;
end
@ -134,7 +134,7 @@ while newRank && t < smpl
Pinf(:,:,t+1) = T*Pinf(:,:,t)*T';
P0=Pinf(:,:,t+1);
if newRank,
newRank = rank(Pinf(:,:,t+1),crit1);
newRank = rank(Pinf(:,:,t+1),diffuse_kalman_tol);
end
if oldRank ~= newRank
disp('univariate_diffuse_kalman_filter:: T does influence the rank of Pinf!')

4
preprocessor/DynareBison.yy
View File

@ -101,7 +101,7 @@ class ParsingDriver;
%token IDENTIFICATION INF_CONSTANT INITVAL INITVAL_FILE BOUNDS JSCALE INIT INFILE INVARS
%token <string_val> INT_NUMBER
%token INV_GAMMA_PDF INV_GAMMA1_PDF INV_GAMMA2_PDF IRF IRF_SHOCKS IRF_PLOT_THRESHOLD IRF_CALIBRATION
%token KALMAN_ALGO KALMAN_TOL SUBSAMPLES OPTIONS TOLF
%token KALMAN_ALGO KALMAN_TOL DIFFUSE_KALMAN_TOL SUBSAMPLES OPTIONS TOLF
%token LAPLACE LIK_ALGO LIK_INIT LINEAR LOAD_IDENT_FILES LOAD_MH_FILE LOAD_PARAMS_AND_STEADY_STATE LOGLINEAR LOGDATA LYAPUNOV
%token LYAPUNOV_FIXED_POINT_TOL LYAPUNOV_DOUBLING_TOL LYAPUNOV_SQUARE_ROOT_SOLVER_TOL LOG_DEFLATOR LOG_TREND_VAR LOG_GROWTH_FACTOR MARKOWITZ MARGINAL_DENSITY MAX MAXIT
%token MFS MH_CONF_SIG MH_DROP MH_INIT_SCALE MH_JSCALE MH_MODE MH_NBLOCKS MH_REPLIC MH_RECOVER POSTERIOR_MAX_SUBSAMPLE_DRAWS MIN MINIMAL_SOLVING_PERIODS
@ -1664,6 +1664,7 @@ estimation_options : o_datafile
| o_filtered_vars
| o_kalman_algo
| o_kalman_tol
| o_diffuse_kalman_tol
| o_xls_sheet
| o_xls_range
| o_filter_step_ahead
@ -2643,6 +2644,7 @@ o_filtered_vars : FILTERED_VARS { driver.option_num("filtered_vars", "1"); };
o_relative_irf : RELATIVE_IRF { driver.option_num("relative_irf", "1"); };
o_kalman_algo : KALMAN_ALGO EQUAL INT_NUMBER { driver.option_num("kalman_algo", $3); };
o_kalman_tol : KALMAN_TOL EQUAL non_negative_number { driver.option_num("kalman_tol", $3); };
o_diffuse_kalman_tol : DIFFUSE_KALMAN_TOL EQUAL non_negative_number { driver.option_num("diffuse_kalman_tol", $3); };
o_marginal_density : MARGINAL_DENSITY EQUAL LAPLACE
{ driver.option_str("mc_marginal_density", "laplace"); }
| MARGINAL_DENSITY EQUAL MODIFIEDHARMONICMEAN

1
preprocessor/DynareFlex.ll
View File

@ -288,6 +288,7 @@ DATE -?[0-9]+([YyAa]|[Mm]([1-9]|1[0-2])|[Qq][1-4]|[Ww]([1-9]{1}|[1-4][0-9]|5[0-2
<DYNARE_STATEMENT>nodiagnostic {return token::NODIAGNOSTIC;}
<DYNARE_STATEMENT>kalman_algo {return token::KALMAN_ALGO;}
<DYNARE_STATEMENT>kalman_tol {return token::KALMAN_TOL;}
<DYNARE_STATEMENT>diffuse_kalman_tol {return token::DIFFUSE_KALMAN_TOL;}
<DYNARE_STATEMENT>forecast {return token::FORECAST;}
<DYNARE_STATEMENT>smoother {return token::SMOOTHER;}
<DYNARE_STATEMENT>bayesian_irf {return token::BAYESIAN_IRF;}