Daniel@0: function bnet = mk_limid(dag, node_sizes, varargin)
Daniel@0: % MK_LIMID Make a limited information influence diagram
Daniel@0: %
Daniel@0: % BNET = MK_LIMID(DAG, NODE_SIZES, ...) 
Daniel@0: % DAG is the adjacency matrix for a directed acyclic graph.
Daniel@0: % The nodes are assumed to be in topological order. Use TOPOLOGICAL_SORT if necessary.
Daniel@0: % For decision nodes, the parents must explicitely include all nodes
Daniel@0: % on which it can depends, in contrast to the implicit no-forgetting assumption of influence diagrams.
Daniel@0: % (For details, see "Representing and solving decision problems with limited information",
Daniel@0: %    Lauritzen and Nilsson, Management Science, 2001.)
Daniel@0: %
Daniel@0: % node_sizes(i) is the number of values node i can take on,
Daniel@0: %   or the length of node i if i is a continuous-valued vector.
Daniel@0: % node_sizes(i) = 1 if i is a utility node.
Daniel@0: % 
Daniel@0: % The list below gives optional arguments [default value in brackets].
Daniel@0: % 
Daniel@0: % chance   - the list of nodes which are random variables [1:N]
Daniel@0: % decision - the list of nodes which are decision nodes [ [] ]
Daniel@0: % utility  - the list of nodes which are utility nodes [ [] ]
Daniel@0: % equiv_class - equiv_class(i)=j  means node i gets its params from CPD{j} [1:N]
Daniel@0: %
Daniel@0: % e.g., limid = mk_limid(dag, ns, 'chance', [1 3], 'utility', [2])
Daniel@0: 
Daniel@0: n = length(dag);
Daniel@0: 
Daniel@0: % default values for parameters
Daniel@0: bnet.chance_nodes = 1:n;
Daniel@0: bnet.equiv_class = 1:n;
Daniel@0: bnet.utility_nodes = [];
Daniel@0: bnet.decision_nodes = [];
Daniel@0: bnet.dnodes = 1:n; % discrete 
Daniel@0: 
Daniel@0: if nargin >= 3
Daniel@0:   args = varargin;
Daniel@0:   nargs = length(args);
Daniel@0:   if ~isstr(args{1})
Daniel@0:     if nargs >= 1, bnet.dnodes = args{1}; end
Daniel@0:     if nargs >= 2, bnet.equiv_class = args{2}; end
Daniel@0:   else    
Daniel@0:     for i=1:2:nargs
Daniel@0:       switch args{i},
Daniel@0:        case 'equiv_class', bnet.equiv_class = args{i+1}; 
Daniel@0:        case 'chance',      bnet.chance_nodes = args{i+1}; 
Daniel@0:        case 'utility',     bnet.utility_nodes = args{i+1}; 
Daniel@0:        case 'decision',    bnet.decision_nodes = args{i+1}; 
Daniel@0:        case 'discrete',    bnet.dnodes = args{i+1}; 
Daniel@0:         otherwise,  
Daniel@0: 	error(['invalid argument name ' args{i}]);       
Daniel@0:       end
Daniel@0:     end
Daniel@0:   end
Daniel@0: end
Daniel@0:  
Daniel@0: bnet.limid = 1;
Daniel@0: 
Daniel@0: bnet.dag = dag;
Daniel@0: bnet.node_sizes = node_sizes(:)';
Daniel@0: 
Daniel@0: bnet.cnodes = mysetdiff(1:n, bnet.dnodes);
Daniel@0: % too many functions refer to cnodes to rename it to cts_nodes - 
Daniel@0: % We hope it won't be confused with chance nodes!
Daniel@0: 
Daniel@0: bnet.parents = cell(1,n);
Daniel@0: for i=1:n
Daniel@0:   bnet.parents{i} = parents(dag, i);
Daniel@0: end
Daniel@0: 
Daniel@0: E = max(bnet.equiv_class);
Daniel@0: mem = cell(1,E);
Daniel@0: for i=1:n
Daniel@0:   e = bnet.equiv_class(i);
Daniel@0:   mem{e} = [mem{e} i];
Daniel@0: end
Daniel@0: bnet.members_of_equiv_class = mem;
Daniel@0: 
Daniel@0: bnet.CPD = cell(1, E);
Daniel@0: 
Daniel@0: % for e=1:E
Daniel@0: %   i = bnet.members_of_equiv_class{e}(1); % pick arbitrary member
Daniel@0: %   switch type{e}
Daniel@0: %     case 'tabular',  bnet.CPD{e} = tabular_CPD(bnet, i);
Daniel@0: %     case 'gaussian', bnet.CPD{e} = gaussian_CPD(bnet, i);
Daniel@0: %     otherwise, error(['unrecognized CPD type ' type{e}]);
Daniel@0: %   end
Daniel@0: % end
Daniel@0: 
Daniel@0: directed = 1;
Daniel@0: if ~acyclic(dag,directed)
Daniel@0:   error('graph must be acyclic')
Daniel@0: end
Daniel@0: 
Daniel@0: bnet.order = topological_sort(bnet.dag);