1function perfect_foresight_solver() 2% Computes deterministic simulations 3% 4% INPUTS 5% None 6% 7% OUTPUTS 8% none 9% 10% ALGORITHM 11% 12% SPECIAL REQUIREMENTS 13% none 14 15% Copyright (C) 1996-2020 Dynare Team 16% 17% This file is part of Dynare. 18% 19% Dynare is free software: you can redistribute it and/or modify 20% it under the terms of the GNU General Public License as published by 21% the Free Software Foundation, either version 3 of the License, or 22% (at your option) any later version. 23% 24% Dynare is distributed in the hope that it will be useful, 25% but WITHOUT ANY WARRANTY; without even the implied warranty of 26% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 27% GNU General Public License for more details. 28% 29% You should have received a copy of the GNU General Public License 30% along with Dynare. If not, see <http://www.gnu.org/licenses/>. 31 32global M_ options_ oo_ 33 34check_input_arguments(options_, M_, oo_); 35 36if isempty(options_.scalv) || options_.scalv == 0 37 options_.scalv = oo_.steady_state; 38end 39 40periods = options_.periods; 41 42options_.scalv= 1; 43 44if options_.debug 45 model_static = str2func([M_.fname,'.static']); 46 for ii=1:size(oo_.exo_simul,1) 47 [residual(:,ii)] = model_static(oo_.steady_state, oo_.exo_simul(ii,:),M_.params); 48 end 49 problematic_periods=find(any(isinf(residual)) | any(isnan(residual)))-M_.maximum_endo_lag; 50 if ~isempty(problematic_periods) 51 period_string=num2str(problematic_periods(1)); 52 for ii=2:length(problematic_periods) 53 period_string=[period_string, ', ', num2str(problematic_periods(ii))]; 54 end 55 fprintf('\n\nWARNING: Value for the exogenous variable(s) in period(s) %s inconsistent with the static model.\n',period_string); 56 fprintf('WARNING: Check for division by 0.\n') 57 end 58end 59 60initperiods = 1:M_.maximum_lag; 61lastperiods = (M_.maximum_lag+periods+1):(M_.maximum_lag+periods+M_.maximum_lead); 62 63oo_ = perfect_foresight_solver_core(M_,options_,oo_); 64 65% If simulation failed try homotopy. 66if ~oo_.deterministic_simulation.status && ~options_.no_homotopy 67 68 if ~options_.noprint 69 fprintf('\nSimulation of the perfect foresight model failed!') 70 fprintf('Switching to a homotopy method...\n') 71 end 72 73 if ~M_.maximum_lag 74 disp('Homotopy not implemented for purely forward models!') 75 disp('Failed to solve the model!') 76 disp('Return with empty oo_.endo_simul.') 77 oo_.endo_simul = []; 78 return 79 end 80 if ~M_.maximum_lead 81 disp('Homotopy not implemented for purely backward models!') 82 disp('Failed to solve the model!') 83 disp('Return with empty oo_.endo_simul.') 84 oo_.endo_simul = []; 85 return 86 end 87 88 % Disable warnings if homotopy 89 warning_old_state = warning; 90 warning off all 91 % Do not print anything 92 oldverbositylevel = options_.verbosity; 93 options_.verbosity = 0; 94 95 % Set initial paths for the endogenous and exogenous variables. 96 endoinit = repmat(oo_.steady_state, 1,M_.maximum_lag+periods+M_.maximum_lead); 97 exoinit = repmat(oo_.exo_steady_state',M_.maximum_lag+periods+M_.maximum_lead,1); 98 99 % Copy the current paths for the exogenous and endogenous variables. 100 exosim = oo_.exo_simul; 101 endosim = oo_.endo_simul; 102 103 current_weight = 0; % Current weight of target point in convex combination. 104 step = .5; % Set default step size. 105 success_counter = 0; 106 iteration = 0; 107 108 if ~options_.noprint 109 fprintf('Iter. \t | Lambda \t | status \t | Max. residual\n') 110 fprintf('++++++++++++++++++++++++++++++++++++++++++++++++++++++++\n') 111 end 112 while (step > options_.dynatol.x) 113 114 if ~isequal(step,1) 115 options_.verbosity = 0; 116 end 117 118 iteration = iteration+1; 119 new_weight = current_weight + step; % Try this weight, and see if it succeeds 120 121 if new_weight >= 1 122 new_weight = 1; % Don't go beyond target point 123 step = new_weight - current_weight; 124 end 125 126 % Compute convex combination for exo path and initial/terminal endo conditions 127 % But take care of not overwriting the computed part of oo_.endo_simul 128 oo_.exo_simul = exosim*new_weight + exoinit*(1-new_weight); 129 oo_.endo_simul(:,[initperiods, lastperiods]) = new_weight*endosim(:,[initperiods, lastperiods])+(1-new_weight)*endoinit(:,[initperiods, lastperiods]); 130 131 % Detect Nans or complex numbers in the solution. 132 path_with_nans = any(any(isnan(oo_.endo_simul))); 133 path_with_cplx = any(any(~isreal(oo_.endo_simul))); 134 135 if isequal(iteration, 1) 136 % First iteration, same initial guess as in the first call to perfect_foresight_solver_core routine. 137 oo_.endo_simul(:,M_.maximum_lag+1:end-M_.maximum_lead) = endoinit(:,1:periods); 138 elseif path_with_nans || path_with_cplx 139 % If solver failed with NaNs or complex number, use previous solution as an initial guess. 140 oo_.endo_simul(:,M_.maximum_lag+1:end-M_.maximum_lead) = saved_endo_simul(:,1+M_.maximum_lag:end-M_.maximum_lead); 141 end 142 143 % Make a copy of the paths. 144 saved_endo_simul = oo_.endo_simul; 145 146 % Solve for the paths of the endogenous variables. 147 [oo_,me] = perfect_foresight_solver_core(M_,options_,oo_); 148 149 if oo_.deterministic_simulation.status == 1 150 current_weight = new_weight; 151 if current_weight >= 1 152 if ~options_.noprint 153 fprintf('%i \t | %1.5f \t | %s \t | %e\n', iteration, new_weight, 'succeeded', me) 154 end 155 break 156 end 157 success_counter = success_counter + 1; 158 if success_counter >= 3 159 success_counter = 0; 160 step = step * 2; 161 end 162 if ~options_.noprint 163 fprintf('%i \t | %1.5f \t | %s \t | %e\n', iteration, new_weight, 'succeeded', me) 164 end 165 else 166 % If solver failed, then go back. 167 oo_.endo_simul = saved_endo_simul; 168 success_counter = 0; 169 step = step / 2; 170 if ~options_.noprint 171 if isreal(me) 172 fprintf('%i \t | %1.5f \t | %s \t | %e\n', iteration, new_weight, 'failed', me) 173 else 174 fprintf('%i \t | %1.5f \t | %s \t | %s\n', iteration, new_weight, 'failed', 'Complex') 175 end 176 end 177 end 178 end 179 if ~options_.noprint 180 fprintf('++++++++++++++++++++++++++++++++++++++++++++++++++++++++\n\n') 181 end 182 options_.verbosity = oldverbositylevel; 183 warning(warning_old_state); 184end 185 186 187if ~isreal(oo_.endo_simul(:)) % cannot happen with bytecode or the perfect_foresight_problem DLL 188 ny = size(oo_.endo_simul, 1) 189 if M_.maximum_lag > 0 190 y0 = real(oo_.endo_simul(:, M_.maximum_lag)); 191 else 192 y0 = NaN(ny, 1); 193 end 194 if M_.maximum_lead > 0 195 yT = real(oo_.endo_simul(:, M_.maximum_lag+periods+1)); 196 else 197 yT = NaN(ny, 1); 198 end 199 yy = real(oo_.endo_simul(:,M_.maximum_lag+(1:periods))); 200 201 residuals = perfect_foresight_problem(yy(:), y0, yT, oo_.exo_simul, M_.params, oo_.steady_state, periods, M_, options_); 202 203 if max(abs(residuals))< options_.dynatol.f 204 oo_.deterministic_simulation.status = 1; 205 oo_.endo_simul=real(oo_.endo_simul); 206 else 207 oo_.deterministic_simulation.status = 0; 208 disp('Simulation terminated with imaginary parts in the residuals or endogenous variables.') 209 end 210end 211 212if oo_.deterministic_simulation.status == 1 213 if ~options_.noprint 214 fprintf('Perfect foresight solution found.\n\n') 215 end 216else 217 fprintf('Failed to solve perfect foresight model\n\n') 218end 219 220dyn2vec(M_, oo_, options_); 221 222if ~isdates(options_.initial_period) && isnan(options_.initial_period) 223 initial_period = dates(1,1); 224else 225 initial_period = options_.initial_period; 226end 227 228ts = dseries(transpose(oo_.endo_simul), initial_period, M_.endo_names); 229assignin('base', 'Simulated_time_series', ts); 230if oo_.deterministic_simulation.status 231 oo_.gui.ran_perfect_foresight = true; 232end 233