function of = a0lhfun(x,s,nobs,nvar,a0indx) % of = a0lhfun(x,s,nobs,nvar,a0indx) -- negative logLH % Note: columns correpond to equations % general program to setup A0 matrix and compute the likelihood % requires % x (parameter vector), % s (covariance matrix of innovations): note already divided by "nobs" % nobs (no of obs), % nvar (no of variables), % a0indx (matrix indicating the free parameters in A0, and each column in A0 corresponds % to an equation) % written by Eric Leeper a0 = zeros(nvar); a0(a0indx) = x; % Note: each column in a0 corresponds to an equation!! % %%ada = chol(a0'*a0); %%ada = log(abs(diag(ada))); %%ada = sum(ada); % ** TZ, 10/15/96, the above two lines can be improved by the following three lines [a0l,a0u] = lu(a0); %ada=diag(abs(a0u)); %ada=sum(log(ada)); ada = sum(log(abs(diag(a0u)))); % %tra = trace(a0'*s*a0); tra = reshape(s,nvar*nvar,1)'*reshape(a0*a0',nvar*nvar,1); %if ada == 0 % of = 1.0; % else % of = -nobs*ada + nobs*.5*tra; %end % commented out by T.Z. for there appears no sense of using "if-end" of = -nobs*ada + nobs*.5*tra;