wolffd@0: function bnet = fgraph_to_bnet(fg)
wolffd@0: % FGRAPH_TO_BNET Convert a factor graph to a Bayes net
wolffd@0: % bnet = fgraph_to_bnet(fg)
wolffd@0: %
wolffd@0: % We assume all factors are tabular_CPD.
wolffd@0: % We create 1 dummy observed node for every factor.
wolffd@0: 
wolffd@0: N = fg.nvars + fg.nfactors;
wolffd@0: vnodes = 1:fg.nvars;
wolffd@0: fnodes = fg.nvars+1:N;
wolffd@0: dag = zeros(N);
wolffd@0: for x=1:fg.nvars
wolffd@0:   dag(x, fnodes(fg.dep{x})) = 1;
wolffd@0: end
wolffd@0: ns = [fg.node_sizes ones(1, fg.nfactors)];
wolffd@0: discrete = [fg.dnodes fnodes];
wolffd@0: bnet = mk_bnet(dag, ns, 'discrete', discrete);
wolffd@0: for x=1:fg.nvars
wolffd@0:   bnet.CPD{x} = tabular_CPD(bnet, x, 'CPT', 'unif');
wolffd@0: end
wolffd@0: ev = cell(1, fg.nvars); % no evidence
wolffd@0: for i=1:fg.nfactors
wolffd@0:   f = fnodes(i);
wolffd@0:   e = fg.equiv_class(i);
wolffd@0:   pot = convert_to_pot(fg.factors{e}, 'd', fg.dom{i}, ev);
wolffd@0:   m = pot_to_marginal(pot);
wolffd@0:   bnet.CPD{f} = tabular_CPD(bnet, f, 'CPT', m.T);
wolffd@0: end
wolffd@0:   
wolffd@0: