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, 48 insertions, 0 deletions

diff --git a/sourcecodes/bnt-master/KPMtools/plot_matrix.m b/sourcecodes/bnt-master/KPMtools/plot_matrix.m
new file mode 100644
index 00000000..58a9a964
--- /dev/null
+++ b/sourcecodes/bnt-master/KPMtools/plot_matrix.m
@@ -0,0 +1,48 @@
+function plot_matrix(G, bw)
+% PLOT_MATRIX Plot a 2D matrix as a grayscale image, and label the axes
+%
+% plot_matrix(M)
+%
+% For 0/1 matrices (eg. adjacency matrices), use
+% plot_matrix(M,1)
+
+if nargin < 2, bw = 0; end
+
+if 0
+  imagesc(G)
+  %image(G)
+  %colormap([1 1 1; 0 0 0]); % black squares on white background
+  %colormap(gray)
+  grid on
+  n = length(G);
+  
+  % shift the grid lines so they don't intersect the squares
+  set(gca,'xtick',1.5:1:n);
+  set(gca,'ytick',1.5:1:n);
+  
+  % Turn off the confusing labels, which are fractional
+  % Ideally we could shift the labels to lie between the axis lines...
+%  set(gca,'xticklabel', []);
+%  set(gca,'yticklabel', []);
+else
+  % solution provided by Jordan Rosenthal <jr@ece.gatech.edu>
+  % You can plot the grid lines manually:
+  % This uses the trick that a point with a value nan does not get plotted.
+  imagesc(G);
+  if bw
+    colormap([1 1 1; 0 0 0]);
+  end
+  n = length(G);
+  x = 1.5:1:n;
+  x = [ x; x; repmat(nan,1,n-1) ];
+  y = [ 0.5 n+0.5 nan ].';
+  y = repmat(y,1,n-1);
+  x = x(:);
+  y = y(:);
+  line(x,y,'linestyle',':','color','k');
+  line(y,x,'linestyle',':','color','k');
+  set(gca,'xtick',1:n)
+  set(gca,'ytick',1:n)
+end
+
+