# one observation at node 4
observations={4:0.0}
# directed edges for kernel BP implementation
edges=model.extract_edges(observations)
print "graph:", graph
print "observations:", observations
print "edges:", edges
# sample data, random number of samples for each edge
n_min=5
n_max=6
data={}
for edge in edges:
samples = model.sample_real(randint(n_min, n_max))
# only sample once per undirected edge
inverse_edge=(edge[1], edge[0])
if (edge not in data and inverse_edge not in data):
data1=samples[edge[0]]
data2=samples[edge[1]]
data[edge]=(data1,data2)
data[inverse_edge]=(data2,data1)
# compute all (here Gaussian) kernels of node data at edges with themselves
kernel=GaussianKernel(sigma=1)
# use the example class for dense matrix data that can be stored in memory
precomputer=PrecomputeDenseMatrixKernelBP(graph, edges, data, observations, \
