wolffd@0: function [engine, loglik] = enter_evidence(engine, evidence, varargin)
wolffd@0: % ENTER_EVIDENCE Add the specified evidence to the network (jtree)
wolffd@0: % [engine, loglik] = enter_evidence(engine, evidence, ...)
wolffd@0: %
wolffd@0: % evidence{i} = [] if X(i) is hidden, and otherwise contains its observed value (scalar or column vector).
wolffd@0: %
wolffd@0: % The following optional arguments can be specified in the form of name/value pairs:
wolffd@0: % [default value in brackets]
wolffd@0: %
wolffd@0: % maximize - if 1, does max-product instead of sum-product [0]
wolffd@0: % soft    - a cell array of soft/virtual evidence;
wolffd@0: %           soft{i} is a prob. distrib. over i's values, or [] [ cell(1,N) ]
wolffd@0: %
wolffd@0: % e.g., engine = enter_evidence(engine, ev, 'soft', soft_ev)
wolffd@0: %
wolffd@0: % For backwards compatibility with BNT2, you can also specify the parameters in the following order
wolffd@0: %  engine = enter_evidence(engine, ev, soft_ev)
wolffd@0: 
wolffd@0: bnet = bnet_from_engine(engine);
wolffd@0: ns = bnet.node_sizes(:);
wolffd@0: N = length(bnet.dag);
wolffd@0: 
wolffd@0: % set default params
wolffd@0: exclude = [];
wolffd@0: soft_evidence = cell(1,N);
wolffd@0: maximize = 0;
wolffd@0: 
wolffd@0: % parse optional params
wolffd@0: args = varargin;
wolffd@0: nargs = length(args);
wolffd@0: if nargs > 0
wolffd@0:   if iscell(args{1})
wolffd@0:     soft_evidence = args{1};
wolffd@0:   else
wolffd@0:     for i=1:2:nargs
wolffd@0:       switch args{i},
wolffd@0:        case 'soft',    soft_evidence = args{i+1}; 
wolffd@0:        case 'maximize', maximize = args{i+1}; 
wolffd@0:        otherwise,  
wolffd@0: 	error(['invalid argument name ' args{i}]);       
wolffd@0:       end
wolffd@0:     end
wolffd@0:   end
wolffd@0: end
wolffd@0: 
wolffd@0: engine.maximize = maximize;
wolffd@0: 
wolffd@0: onodes = find(~isemptycell(evidence));
wolffd@0: hnodes = find(isemptycell(evidence));
wolffd@0: pot_type = determine_pot_type(bnet, onodes);
wolffd@0:  if strcmp(pot_type, 'cg')
wolffd@0:   check_for_cd_arcs(onodes, bnet.cnodes, bnet.dag);
wolffd@0: end
wolffd@0: 
wolffd@0: 
wolffd@0: hard_nodes = 1:N;
wolffd@0: soft_nodes = find(~isemptycell(soft_evidence));
wolffd@0: S = length(soft_nodes);
wolffd@0: if S > 0
wolffd@0:   assert(pot_type == 'd');
wolffd@0:   assert(mysubset(soft_nodes, bnet.dnodes));
wolffd@0: end
wolffd@0:  
wolffd@0: % Evaluate CPDs with evidence, and convert to potentials  
wolffd@0: pot = cell(1, N+S);
wolffd@0: for n=1:N
wolffd@0:   fam = family(bnet.dag, n);
wolffd@0:   e = bnet.equiv_class(n);
wolffd@0:   pot{n} = convert_to_pot(bnet.CPD{e}, pot_type, fam(:), evidence);
wolffd@0: end
wolffd@0: 
wolffd@0: for i=1:S
wolffd@0:   n = soft_nodes(i);
wolffd@0:   pot{N+i} = dpot(n, ns(n), soft_evidence{n});
wolffd@0: end
wolffd@0: 
wolffd@0: %clqs = engine.clq_ass_to_node([hard_nodes soft_nodes]); 
wolffd@0: %[clpot, loglik] = enter_soft_evidence(engine, clqs, pot, onodes, pot_type);
wolffd@0: %engine.clpot = clpot; % save the results for marginal_nodes
wolffd@0: 
wolffd@0: 
wolffd@0: clique = engine.clq_ass_to_node([hard_nodes soft_nodes]); 
wolffd@0: potential = pot;
wolffd@0: 
wolffd@0: 
wolffd@0: % Set the clique potentials to all 1s
wolffd@0: C = length(engine.cliques);
wolffd@0: for i=1:C
wolffd@0:   engine.clpot{i} = mk_initial_pot(pot_type, engine.cliques{i}, ns, bnet.cnodes, onodes);
wolffd@0: end
wolffd@0: 
wolffd@0: % Multiply on specified potentials
wolffd@0: for i=1:length(clique)
wolffd@0:   c = clique(i);
wolffd@0:   engine.clpot{c} = multiply_by_pot(engine.clpot{c}, potential{i});
wolffd@0: end
wolffd@0: 
wolffd@0: root = 1; % arbitrary
wolffd@0: engine = collect_evidence(engine, root);
wolffd@0: engine = distribute_evidence(engine, root);
wolffd@0: 
wolffd@0: ll = zeros(1, C);
wolffd@0: for i=1:C
wolffd@0:   [engine.clpot{i}, ll(i)] = normalize_pot(engine.clpot{i});
wolffd@0: end
wolffd@0: loglik = ll(1); % we can extract the likelihood from any clique
wolffd@0: