2019-12-21 10:44:23 +01:00
|
|
|
function [LIK,lik] = nonlinear_kalman_filter(ReducedForm, Y, start, ParticleOptions, ThreadsOptions, DynareOptions, Model)
|
|
|
|
|
2015-10-09 14:42:53 +02:00
|
|
|
% Evaluates the likelihood of a non-linear model approximating the
|
|
|
|
% predictive (prior) and filtered (posterior) densities for state variables
|
|
|
|
% by a Kalman filter.
|
|
|
|
% Gaussian distribution approximation is done by:
|
|
|
|
% - a spherical-radial cubature (ref: Arasaratnam & Haykin, 2009).
|
|
|
|
% - a scaled unscented transform cubature (ref: Julier & Uhlmann 1995)
|
|
|
|
% - Monte-Carlo draws from a multivariate gaussian distribution.
|
|
|
|
% First and second moments of prior and posterior state densities are computed
|
|
|
|
% from the resulting nodes/particles and allows to generate new distributions at the
|
|
|
|
% following observation.
|
|
|
|
% Pros: The use of nodes is much faster than Monte-Carlo Gaussian particle and standard particles
|
2017-05-18 23:59:10 +02:00
|
|
|
% filters since it treats a lesser number of particles.
|
|
|
|
% Cons: 1. Application a linear projection formulae in a nonlinear context.
|
|
|
|
% 2. Parameter estimations may be biaised if the model is truly non-gaussian since predictive and
|
2015-10-09 14:42:53 +02:00
|
|
|
% filtered densities are unimodal.
|
|
|
|
%
|
|
|
|
% INPUTS
|
|
|
|
% Reduced_Form [structure] Matlab's structure describing the reduced form model.
|
|
|
|
% Y [double] matrix of original observed variables.
|
|
|
|
% start [double] structural parameters.
|
|
|
|
% ParticleOptions [structure] Matlab's structure describing options concerning particle filtering.
|
|
|
|
% ThreadsOptions [structure] Matlab's structure.
|
|
|
|
%
|
|
|
|
% OUTPUTS
|
|
|
|
% LIK [double] scalar, likelihood
|
|
|
|
% lik [double] vector, density of observations in each period.
|
|
|
|
%
|
|
|
|
% REFERENCES
|
|
|
|
%
|
|
|
|
% NOTES
|
|
|
|
% The vector "lik" is used to evaluate the jacobian of the likelihood.
|
2019-12-21 10:44:23 +01:00
|
|
|
|
2022-04-13 14:47:52 +02:00
|
|
|
% Copyright © 2009-2022 Dynare Team
|
2015-10-09 14:42:53 +02:00
|
|
|
%
|
|
|
|
% 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
|
2021-06-09 17:21:49 +02:00
|
|
|
% along with Dynare. If not, see <https://www.gnu.org/licenses/>.
|
2015-10-09 14:42:53 +02:00
|
|
|
|
|
|
|
% Set default
|
|
|
|
if isempty(start)
|
|
|
|
start = 1;
|
|
|
|
end
|
|
|
|
|
2022-09-19 15:43:10 +02:00
|
|
|
order = DynareOptions.order;
|
|
|
|
|
2019-12-21 10:44:23 +01:00
|
|
|
if ReducedForm.use_k_order_solver
|
|
|
|
dr = ReducedForm.dr;
|
2022-02-04 11:29:31 +01:00
|
|
|
udr = ReducedForm.udr;
|
2019-12-21 10:44:23 +01:00
|
|
|
else
|
|
|
|
% Set local state space model (first-order approximation).
|
|
|
|
ghx = ReducedForm.ghx;
|
|
|
|
ghu = ReducedForm.ghu;
|
|
|
|
% Set local state space model (second-order approximation).
|
|
|
|
ghxx = ReducedForm.ghxx;
|
|
|
|
ghuu = ReducedForm.ghuu;
|
|
|
|
ghxu = ReducedForm.ghxu;
|
2023-07-01 16:43:48 +02:00
|
|
|
ghs2 = ReducedForm.ghs2;
|
2022-09-19 15:43:10 +02:00
|
|
|
if (order == 3)
|
|
|
|
% Set local state space model (third order approximation).
|
|
|
|
ghxxx = ReducedForm.ghxxx;
|
|
|
|
ghuuu = ReducedForm.ghuuu;
|
|
|
|
ghxxu = ReducedForm.ghxxu;
|
|
|
|
ghxuu = ReducedForm.ghxuu;
|
|
|
|
ghxss = ReducedForm.ghxss;
|
|
|
|
ghuss = ReducedForm.ghuss;
|
|
|
|
end
|
2015-10-09 14:42:53 +02:00
|
|
|
end
|
|
|
|
|
|
|
|
constant = ReducedForm.constant;
|
2023-07-01 16:43:48 +02:00
|
|
|
steadystate = ReducedForm.steadystate;
|
2015-10-09 14:42:53 +02:00
|
|
|
state_variables_steady_state = ReducedForm.state_variables_steady_state;
|
|
|
|
|
2019-12-21 10:44:23 +01:00
|
|
|
mf0 = ReducedForm.mf0;
|
|
|
|
mf1 = ReducedForm.mf1;
|
|
|
|
sample_size = size(Y,2);
|
|
|
|
number_of_state_variables = length(mf0);
|
|
|
|
number_of_observed_variables = length(mf1);
|
|
|
|
number_of_structural_innovations = length(ReducedForm.Q);
|
2015-10-09 14:42:53 +02:00
|
|
|
|
|
|
|
% compute gaussian quadrature nodes and weights on states and shocks
|
2018-06-08 20:58:48 +02:00
|
|
|
if ParticleOptions.proposal_approximation.montecarlo
|
2019-12-21 10:44:23 +01:00
|
|
|
nodes = randn(ParticleOptions.number_of_particles,number_of_state_variables+number_of_structural_innovations);
|
|
|
|
weights = 1/ParticleOptions.number_of_particles;
|
|
|
|
weights_c = weights;
|
2018-06-08 20:58:48 +02:00
|
|
|
elseif ParticleOptions.proposal_approximation.cubature
|
2019-12-21 10:44:23 +01:00
|
|
|
[nodes,weights] = spherical_radial_sigma_points(number_of_state_variables+number_of_structural_innovations);
|
|
|
|
weights_c = weights;
|
2015-10-09 14:42:53 +02:00
|
|
|
elseif ParticleOptions.proposal_approximation.unscented
|
|
|
|
[nodes,weights,weights_c] = unscented_sigma_points(number_of_state_variables+number_of_structural_innovations,ParticleOptions);
|
|
|
|
else
|
|
|
|
error('Estimation: This approximation for the proposal is not implemented or unknown!')
|
|
|
|
end
|
|
|
|
|
|
|
|
if ParticleOptions.distribution_approximation.montecarlo
|
2023-09-25 13:25:39 +02:00
|
|
|
DynareOptions=set_dynare_seed_local_options(DynareOptions,'default');
|
2017-05-18 23:59:10 +02:00
|
|
|
end
|
2015-10-09 14:42:53 +02:00
|
|
|
|
|
|
|
% Get covariance matrices
|
|
|
|
H = ReducedForm.H;
|
2017-05-18 23:59:10 +02:00
|
|
|
H_lower_triangular_cholesky = chol(H)' ;
|
|
|
|
Q_lower_triangular_cholesky = chol(ReducedForm.Q)';
|
2015-10-09 14:42:53 +02:00
|
|
|
|
|
|
|
% Get initial condition for the state vector.
|
|
|
|
StateVectorMean = ReducedForm.StateVectorMean;
|
|
|
|
StateVectorVarianceSquareRoot = chol(ReducedForm.StateVectorVariance)';
|
|
|
|
|
|
|
|
% Initialization of the likelihood.
|
|
|
|
lik = NaN(sample_size,1);
|
|
|
|
LIK = NaN;
|
|
|
|
|
|
|
|
for t=1:sample_size
|
|
|
|
xbar = [StateVectorMean ; zeros(number_of_structural_innovations,1) ] ;
|
2019-12-21 10:44:23 +01:00
|
|
|
sqr_Px = [StateVectorVarianceSquareRoot zeros(number_of_state_variables,number_of_structural_innovations);
|
|
|
|
zeros(number_of_structural_innovations,number_of_state_variables) Q_lower_triangular_cholesky];
|
2015-10-09 14:42:53 +02:00
|
|
|
sigma_points = bsxfun(@plus,xbar,sqr_Px*(nodes'));
|
|
|
|
StateVectors = sigma_points(1:number_of_state_variables,:);
|
|
|
|
epsilon = sigma_points(number_of_state_variables+1:number_of_state_variables+number_of_structural_innovations,:);
|
|
|
|
yhat = bsxfun(@minus,StateVectors,state_variables_steady_state);
|
2019-12-21 10:44:23 +01:00
|
|
|
if ReducedForm.use_k_order_solver
|
2022-02-04 11:29:31 +01:00
|
|
|
tmp = local_state_space_iteration_k(yhat, epsilon, dr, Model, DynareOptions, udr);
|
2019-12-21 10:44:23 +01:00
|
|
|
else
|
2022-09-19 15:43:10 +02:00
|
|
|
if order == 2
|
|
|
|
tmp = local_state_space_iteration_2(yhat, epsilon, ghx, ghu, constant, ghxx, ghuu, ghxu, ThreadsOptions.local_state_space_iteration_2);
|
|
|
|
elseif order == 3
|
2023-07-01 16:43:48 +02:00
|
|
|
tmp = local_state_space_iteration_3(yhat, epsilon, ghx, ghu, ghxx, ghuu, ghxu, ghs2, ghxxx, ghuuu, ghxxu, ghxuu, ghxss, ghuss, steadystate, ThreadsOptions.local_state_space_iteration_3, false);
|
2022-09-19 15:43:10 +02:00
|
|
|
end
|
2019-12-21 10:44:23 +01:00
|
|
|
end
|
2015-10-09 14:42:53 +02:00
|
|
|
PredictedStateMean = tmp(mf0,:)*weights ;
|
|
|
|
PredictedObservedMean = tmp(mf1,:)*weights;
|
|
|
|
if ParticleOptions.proposal_approximation.cubature || ParticleOptions.proposal_approximation.montecarlo
|
|
|
|
PredictedStateMean = sum(PredictedStateMean,2);
|
|
|
|
PredictedObservedMean = sum(PredictedObservedMean,2);
|
|
|
|
dState = bsxfun(@minus,tmp(mf0,:),PredictedStateMean)'.*sqrt(weights);
|
|
|
|
dObserved = bsxfun(@minus,tmp(mf1,:),PredictedObservedMean)'.*sqrt(weights);
|
|
|
|
big_mat = [dObserved dState ; [H_lower_triangular_cholesky zeros(number_of_observed_variables,number_of_state_variables)] ];
|
2019-12-21 10:44:23 +01:00
|
|
|
[~, mat] = qr2(big_mat,0);
|
2015-10-09 14:42:53 +02:00
|
|
|
mat = mat';
|
|
|
|
PredictedObservedVarianceSquareRoot = mat(1:number_of_observed_variables,1:number_of_observed_variables);
|
|
|
|
CovarianceObservedStateSquareRoot = mat(number_of_observed_variables+(1:number_of_state_variables),1:number_of_observed_variables);
|
|
|
|
StateVectorVarianceSquareRoot = mat(number_of_observed_variables+(1:number_of_state_variables),number_of_observed_variables+(1:number_of_state_variables));
|
|
|
|
PredictionError = Y(:,t) - PredictedObservedMean;
|
|
|
|
StateVectorMean = PredictedStateMean + (CovarianceObservedStateSquareRoot/PredictedObservedVarianceSquareRoot)*PredictionError;
|
|
|
|
else
|
|
|
|
dState = bsxfun(@minus,tmp(mf0,:),PredictedStateMean);
|
|
|
|
dObserved = bsxfun(@minus,tmp(mf1,:),PredictedObservedMean);
|
|
|
|
PredictedStateVariance = dState*diag(weights_c)*dState';
|
|
|
|
PredictedObservedVariance = dObserved*diag(weights_c)*dObserved' + H;
|
|
|
|
PredictedStateAndObservedCovariance = dState*diag(weights_c)*dObserved';
|
|
|
|
PredictionError = Y(:,t) - PredictedObservedMean;
|
|
|
|
KalmanFilterGain = PredictedStateAndObservedCovariance/PredictedObservedVariance;
|
|
|
|
StateVectorMean = PredictedStateMean + KalmanFilterGain*PredictionError;
|
|
|
|
StateVectorVariance = PredictedStateVariance - KalmanFilterGain*PredictedObservedVariance*KalmanFilterGain';
|
2017-01-23 12:20:24 +01:00
|
|
|
[StateVectorVarianceSquareRoot, p]= chol(StateVectorVariance,'lower');
|
|
|
|
if p
|
|
|
|
LIK=-Inf;
|
|
|
|
lik(t)=-Inf;
|
|
|
|
return
|
|
|
|
end
|
2019-12-21 10:44:23 +01:00
|
|
|
[~, p]= chol(PredictedObservedVariance,'lower');
|
2017-01-23 12:20:24 +01:00
|
|
|
if p
|
|
|
|
LIK=-Inf;
|
|
|
|
lik(t)=-Inf;
|
|
|
|
return
|
|
|
|
end
|
2015-10-09 14:42:53 +02:00
|
|
|
end
|
2018-06-28 09:20:19 +02:00
|
|
|
lik(t) = log( sum(probability2(Y(:,t),H_lower_triangular_cholesky,tmp(mf1,:)).*weights,1) ) ;
|
2015-10-09 14:42:53 +02:00
|
|
|
end
|
|
|
|
|
2022-02-04 11:29:31 +01:00
|
|
|
LIK = -sum(lik(start:end));
|