function [engine, newbel, loglik] = dbn_update(engine, oldbel, evidence)% DBN_UPDATE Do one step of Bayesian updating in the dynamic model% [engine, newbel, loglik] = dbn_update(engine, oldbel, evidence)%% oldbel is Pr(X(t-1)|Y(1:t-1)), newbel is Pr(X(t)|Y(1:t))% evidence{i} contains the evidence on node i in slice tss = length(evidence);bnet = bnet_from_engine(engine);CPDpot = cell(1, ss);for n=1:ss  fam = family(bnet.dag, n, 2);  e = bnet.equiv_class(n, 2);  CPDpot{n} = CPD_to_table(bnet.CPD{e}, fam, bnet.node_sizes, bnet.cnodes, evidence);endclqs = [engine.fwd.clq_ass_to_cluster(:,1); engine.fwd.clq_ass_to_node(:,2)];ndx = [engine.fwd.cluster_ndx(:,1); engine.fwd.CPD_ndx(:,2)];pots = [oldbel; CPDpot];[newbel, loglik] = enter_soft_evidence(engine.fwd.engine, clqs, pots, ndx);