BNW using Octave instead of Matlab.

This version of BNW should perform the same as the original version. The only difference is that it uses Octave instead of Matlab when running BayesNet Toolbox during parameter learning.

I am calling this BNW_1.02. It can be accessed at:
compbio.uthsc.edu/BNW_1.02

1 files changed, 108 insertions, 0 deletions

diff --git a/sourcecodes/bnt-master/BNT/test_BNT.m b/sourcecodes/bnt-master/BNT/test_BNT.m
new file mode 100644
index 00000000..0db86e2e
--- /dev/null
+++ b/sourcecodes/bnt-master/BNT/test_BNT.m
@@ -0,0 +1,108 @@
+% If all is well, all of these scripts should run without errors.
+
+
+% bnets
+cg1
+cg2
+discrete1
+fa1
+gaussian1
+gaussian2
+if exist('@gibbs_sampling_inf_engine/private/compute_posterior','file')
+  % only exists if installC has been run
+  gibbs_test1
+end
+learn1
+lw1
+mfa1
+mixexp1
+mixexp2
+mixexp3
+mog1
+mpe1
+mpe2
+qmr1
+qmr2
+sample1
+softev1
+softmax1
+sprinkler1
+
+
+% belief propagation
+belprop_polytree_discrete
+belprop_polytree_gauss % alag
+belprop_polytree_cg
+belprop_loop1_discrete
+belprop_loop1_gauss
+belprop_loopy_discrete
+belprop_loopy_gauss
+belprop_loopy_cg % like cg1
+
+
+% factor graphs
+%fg1   failed since marginals were not exact
+
+fg2
+fg3
+fg_mrf1
+fg_mrf2
+
+
+% Structure learning
+bic1
+cooper_yoo
+k2demo1
+mcmc1
+model_select1
+pc1
+%pc2   failed due to numerical problems in KPMstats/cond_indep_fisher_z
+
+
+
+
+% limids
+asia_dt1
+id1
+oil1
+pigs1
+
+
+% dbns
+arhmm1
+bat1
+bkff1
+chmm1
+dhmm1
+filter_test1
+ghmm1
+kalman1
+kjaerulff1
+loopy_dbn1
+mhmm1
+mildew1
+reveal1
+viterbi1
+water1
+
+
+% HHMMs
+abcd_hhmm
+sample_square_hhmm_discrete
+%learn_square_hhmm_cts
+sample_motif_hhmm
+
+%sparse jtree engine & ndx 2TBN engine
+if exist('@jtree_sparse_inf_engine/init_pot','file')
+  % only exists if installC has been run
+  discrete2
+  discrete3 
+  filter_test1
+  water2
+end
+
+%find . -path '*.m' -exec wc -l {} \; | ~/count.pl
+
+% we cannot use tic;toc to time test_BNT, since functions within this script
+% reset the tic;toc timer. Hence we use the following:
+%clock0=clock; cpu0 = cputime; test_BNT; cpu=cputime-cpu0; elapsed=etime(clock, clock0)