49 lines
1.1 KiB
Matlab
49 lines
1.1 KiB
Matlab
|
function marginal = marginal_density(x2,logpost2)
|
||
|
|
global M_ options_
|
||
|
|
|
||
|
|
nruns = size(x2,1);
|
||
|
|
nblck = size(x2,3);
|
||
|
|
npara = size(x2,2);
|
||
|
|
|
||
|
|
B = 1;
|
||
|
|
T = nruns-round(options_.mh_drop*nruns)+1;
|
||
|
|
N = T*nblck;
|
||
|
|
simulations = zeros(N,npara);
|
||
|
|
lposterior = zeros(N,1);
|
||
|
|
|
||
|
|
while B <= nblck;
|
||
|
|
simulations((B-1)*T+1:B*T,:) = x2(round(options_.mh_drop*nruns):nruns,:,B);
|
||
|
|
lposterior((B-1)*T+1:B*T) = logpost2(round(options_.mh_drop*nruns):nruns,B);
|
||
|
|
B = B + 1;
|
||
|
|
end;
|
||
|
|
|
||
|
|
lpost_mode = max(lposterior);
|
||
|
|
|
||
|
|
MU = mean(simulations)';
|
||
|
|
SIGMA = zeros(npara,npara);
|
||
|
|
for i=1:N;
|
||
|
|
SIGMA = SIGMA + (simulations(i,:)'-MU)*(simulations(i,:)'-MU)';
|
||
|
|
end;
|
||
|
|
SIGMA = SIGMA/N;
|
||
|
|
|
||
|
|
DetSIGMA = det(SIGMA);
|
||
|
|
InvSIGMA = inv(SIGMA);
|
||
|
|
marginal = [];
|
||
|
|
|
||
|
|
for p = 0.1:0.1:0.9;
|
||
|
|
critval = qchisq(p,npara);
|
||
|
|
tmp = 0;
|
||
|
|
for i = 1:N;
|
||
|
|
deviation = (simulations(i,:)-MU')*InvSIGMA*((simulations(i,:)-MU'))';
|
||
|
|
if deviation <= critval;
|
||
|
|
lftheta = -log(p)-(npara*log(2*pi)+log(DetSIGMA)+deviation)/2;
|
||
|
|
tmp = tmp + exp(lftheta - lposterior(i)+lpost_mode);
|
||
|
|
end;
|
||
|
|
end;
|
||
|
|
marginal = cat(1,marginal,[p,lpost_mode-log(tmp/N)]);
|
||
|
|
end;
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
|