-
Notifications
You must be signed in to change notification settings - Fork 64
/
spm_DEM_generate.m
104 lines (95 loc) · 2.81 KB
/
spm_DEM_generate.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
function [DEM] = spm_DEM_generate(M,U,P,h,g)
% Generates data for a Hierarchical Dynamic Model (HDM)
% FORMAT [DEM] = spm_DEM_generate(M,N,P,h,g): N-samples using z
% FORMAT [DEM] = spm_DEM_generate(M,U,P,h,g): size(U,2) samples using U
%
% M(i) - HDM
% U(n x N} - causes or N number of causes
% P{i} - model-parameters for level i (defaults to M.pE)
% h{i} - log-precisions for level i (defaults to 32 - no noise)
% g{i} - log-precisions for level i (defaults to 32 - no noise)
%
% generates
% DEM.M - hierarchical model (checked)
% DEM.Y - responses or data
%
% and true causes NB: v{end} = U or z{end} (last level innovations)
% DEM.pU.v
% DEM.pU.x
% DEM.pU.e
% DEM.pP.P
% DEM.pH.h
%__________________________________________________________________________
% Copyright (C) 2008 Wellcome Trust Centre for Neuroimaging
% Karl Friston
% $Id: spm_DEM_generate.m 5219 2013-01-29 17:07:07Z spm $
% sequence length specified by priors on causes
%--------------------------------------------------------------------------
M = spm_DEM_M_set(M);
DEM.M = M;
try
if length(U) > 1
N = size(U,2);
else
N = U;
U = sparse(M(end).l,N);
end
catch
warndlg('Please specify model inputs U or number of samples')
return
end
% initialize model-parameters if specified
%--------------------------------------------------------------------------
try, P; if ~iscell(P), P = {P}; end, catch, P = {M.pE}; end
try, h; if ~iscell(h), h = {h}; end, catch, h = {}; end
try, g; if ~iscell(P), g = {g}; end, catch, g = {}; end
% transcribe parameters and hyperparameters into prior expectations
%--------------------------------------------------------------------------
m = length(M);
for i = 1:m
try
M(i).pE = spm_unvec(P{i},M(i).pE);
catch
try
M(i).pE = P{i};
end
end
end
for i = 1:m
try
M(i).hE = h{i};
catch
M(i).hE = (M(i).hE - M(i).hE) + 32;
end
end
for i = 1:m
try
M(i).gE = g{i};
catch
M(i).gE = (M(i).gE - M(i).gE) + 32;
end
end
% re-set M and create innovations
%--------------------------------------------------------------------------
M = spm_DEM_M_set(M);
DEM.G = M;
[z,w] = spm_DEM_z(M,N);
% place exogenous causes in cell array
%--------------------------------------------------------------------------
for i = 1:m
u{i} = sparse(M(i).l,N);
end
u{m} = U;
% integrate HDM to obtain causal (v) and hidden states (x)
%--------------------------------------------------------------------------
[v,x,z,w] = spm_DEM_int(M,z,w,u);
% Fill in DEM with response and its causes
%--------------------------------------------------------------------------
DEM.Y = v{1};
DEM.pU.v = v;
DEM.pU.x = x;
DEM.pU.z = z;
DEM.pU.w = w;
DEM.pP.P = {M.pE};
DEM.pH.h = {M.hE};
DEM.pH.g = {M.gE};