max_marginalization.m

The BNL toolbox is a set of Matlab functions for defining and estimating the parameters of a Bayesi

function [max_configs, degeneracy]=find_max_configs(parms,design,parents,node_sizes...
    ,equiv_class,equiv_class_time,terminal_merged_nodes,...
    preorder, postorder,cliquetable, septable,pot_to_CPT,N)    



%E step

lin_pred=construct_lin_pred(parms,design,parents,node_sizes,equiv_class,equiv_class_time,terminal_merged_nodes,N);

equiv_class_CPTs=construct_equiv_class_CPT(lin_pred,N);

bigCPTs=construct_bigCPTs(equiv_class_CPTs,equiv_class,...
    evidence_nodes,partial_evidence_nodes,terminal_merged_nodes,hid_nodes,...
    gausskwadnodes,N);

[cliquetable, septable]=franks_init_pot(cliquetable,septable,bigCPTs,parents);
%max_marginaliation
%collect
[septable,cliquetable]=max_propagate_messages(postorder,septable, cliquetable);%kan sneller!
%distribute
[septable,cliquetable]=max_propagate_messages(preorder,septable, cliquetable);
        
%find max configuration 
[cliquetable,septable]=select_max_config(cliquetable,septable);

%look for degeneracies
%collect
[septable,cliquetable,schaal_collect]=propagate_messages(postorder,septable, cliquetable);
%distribute
[septable,max_configs,schaal_distribute]=propagate_messages(preorder,septable, cliquetable);
        
%compute degeneracies

schaal=schaal_collect+schaal_distribute;
degeneracy=loglik(max_configs.pots{1},schaal);