cfp
November 15, 2010, 1:39pm
1
The current latest code seems to omit the “ys” parameter when calling *_steadystate in evaluate_likelihood. Is this deliberate?
Passing oo_.steady_state is useful when there’s no analytic solution since it gives a base from which to start fsolve.
Tom
cfp
November 15, 2010, 2:59pm
2
What would be particularly useful would be if they were all called with the return of the *_steadystate function from the previous call, rather than the auxiliary variable expanded version.
cfp
November 15, 2010, 3:04pm
3
E.g.:
if ~isfield( oo_, 'reduced_steady_state' )
if isfield( M_, 'orig_endo_nbr' )
NumberOfEndogenousVariables = M_.orig_endo_nbr;
else
NumberOfEndogenousVariables = M_.endo_nbr;
end
oo_.reduced_steady_state = zeros( NumberOfEndogenousVariables );
end
[ys,tchek] = feval([M_.fname '_steadystate'],oo_.reduced_steady_state,...
[zeros(M_.exo_nbr,1);...
oo_.exo_det_steady_state]);
oo_.reduced_steady_state = ys;